An Assessment of Image Quality in Geology Works from the HathiTrust Digital Library

Abstract

This study assesses the quality of both images and text in a sample from the 2,180 works on geology from the HathiTrust Digital Library (multi-institutional digital repository)--an outgrowth of the Michigan Digitization Project and partnership with Google, Inc. A random sample of 180 (consisting of 47,287 pages) was made and reveals many patterns and characteristics of the digital manifestations of these works. The good news is that of the total 47,287 pages that were reviewed, only 2.5% had errors. The bad news, of the 180 works, 114 or 63% had at least one scanning error. It is important for librarians and readers to know the strengths and shortcomings of this repository in considering future decisions on both de-accessioning and remote storage of works from libraries.