This month JSTOR released an "updated" version of the archived back issues of Science. First converted to electronic format by JSTOR in 1999, Science is among the largest and more complex journals ever digitized and archived by JSTOR. This release of the new version is one of the first visible work products of our large-scale retrospective conversion ('retro') initiatives—a series of efforts designed to bring journals up to our current specifications.

Science began in 1880 and, over 120 years, has published more than 380,000 pages. The first task in the retro project was daunting—the paper volumes of Science that JSTOR originally scanned were on loan and had been returned to their owners post-digitization, requiring the JSTOR Production unit to reassemble a replacement back run. This took more than a year, and the volumes came from dozens of sources. Once a full set of volumes was secured, a thorough review revealed some minor differences between the volumes originally digitized and the present set. Discrepancies occurred primarily in advertisement pages, which are in some cases removed by libraries in the volumes they store. Each discrepancy was assessed, documented, and resolved.

Volumes were then shipped along with detailed instructions prepared by JSTOR to our scanning partner. The focus of their activities centered on the addition of image compositing, Unicode encoding, ASCII transliterations, and the capture of illustration captions. [For an explanation of these items see: http://www.jstor.org/news/2005.03/retro.html.] Every title, author name, and abstract were examined for non-ASCII characters, such as Arabic or Greek, which were then Unicode-encoded and transliterated according to accepted standards. Each page was reviewed for potential image and caption information. In total, 153,930 grayscale and color illustrations were rescanned in TIFF format and enabled for image compositing. JSTOR worked closely with our partner throughout, and data returned to JSTOR underwent a thorough review and quality control process.

Once the new data was in hand, we faced another challenge: replacing material already released and under continuous use by researchers and students. Working with our production team, we ensured there was no interruption or lag time in service. Through a modified version of our normal data release process, we seamlessly added the new data. Researchers and students accessed the old content until the very moment the reworked content replaced it, without any gap or delay.

We are very pleased to introduce Science, again, and hope you will take a fresh look. Additionally, "updated" versions of sixteen Language & Literature titles were released into the archive without interruption in September. One of these journals, Shakespeare Quarterly, features reworked illustrations from the seventeenth and eighteenth centuries, including the first American engraving of Shakespeare in 1787 (Volume 12, No. 2, Spring 1961).

Work on these titles has proven a valuable testing ground for developing the processes and tools needed to manage a continuous workflow of retro projects. This activity is essential to ensuring that the twenty million page archive keeps pace with evolving preservation, data capture, and online delivery standards, and we look forward to introducing titles again, and again, in the years ahead.

Last updated on September 8, 2006