We seem to have had rather bad luck with visitor counters, which operate by visitors passing through a beam, each pass augmenting an electronic counter by one. Various models which we have tried seem to have produced very erratic, not to say sometimes impossible, results. (We are of course aware that a percentage of error has to be expected as a consequence of visitors entering together so that the beam does not see any gap between them). Could anyone offer: a) info on models which they have found reliable b) views on what % error should be taken as par for the course c) views on what improvement in accuracy is likely to be achieved by substituting these counters for the process of manual counting in sample weeks. Do they produce more accurate results, or do they simply offer a cheaper way of generating the same level of error? I shall be happy to circulate a summary of responses to this query. Peter Marshall Project Development Librarian Bexley Council Thamesmead Centre Yarnton Way Erith DA18 4DR Tel: 020 8320 4135 (Direct Line) email: [log in to unmask]