This investigation is a large scale study of the real world Web information seeking activity of 1,050 full-time undergraduates studying at a UK higher education institution. The study takes the form of a transaction log analysis of a Web log which records over a two year period all the 1,990,488 URLs requested by the students during 46,558 daily sessions. The analysis focuses on how individual students seek Web information. This is made possible by each user being (anonymously) identified throughout the Web log. Both longitudinal and non-longitudinal or repeat study analyses are undertaken. The analyses make use of a novel session-conformance metric which measures the similarity/dissimilarity of the collection of Website requests made during each session. Over time student-users become more individually distinctive in respect of their 'Web territories' or the collections of Websites which they visit and revisit during each session. Student-users become more territorial in that they increasingly locate their Web information resources from within their own Web territories. 'Searching' occurs in only half of all sessions and student-users undertake less 'searching' as their Web territories become more strongly developed. These findings are interpreted using the notion of a personal Web information infrastructure which is based on Marchionini's idea of a personal information infrastructure (Marchionini, 1995). A student-user's personal Web information infrastructure is represented by his (or her) territory. As student-users become more proficient at locating Web information resources to satisfy their individual information needs so they build or strengthen their personal Web information infrastructures.