Dr. Elizabeth Pan argues that rank lists of journals based on citations are reliable indicators-and by implication useful indicators—of journal usage in libraries [12]. I would argue that, on the contrary, no measure of journal use other than one derived from a local-use study is of any significant practical valueto libraries. In restating this, I fear it is necessary to repeat what has been said in several previously published papers [5, 6, 9]. Scales reported a low correlation between ISl’s list ofjournals in order of citations received and the list of journals in order of inter-library loan demands received by the British Library Lending Division [14]. Dr. Pan attributes this to the fact that the demand received by the British Library Lending Division is marginal (and there-fore not as representative as the ISI citation data). This argument is in fact dealt with in Scales’s article, and more fully in published cor-respondence following the paper [2, 7, 8, ll, 13, 15, 16, 17]. The essential point is that over a third of all the interlibrary loan demand for journals received by the British Library Lending Division comes from smaller specialized libraries, many of whose requests are for core journals: what is core to one library is marginal to another. A glance at the list of the most requested journals shows that they are all high-status, high-use, commonly held journals [1, 10]. Newcastle University Library found that demand on the British Library Lend-