Stages V and VI: Final Camera-Ready Production
This is basically the last run of stage IV. However, rather than put the DVI files on the Laserjet they are sent to a high-quality typesetter. On the PC the DVI files are processed and converted to a Postscript file which are then transferred back to the VAX for printing on bromide on the OUCS Prism printer. The final camera-ready copy is then passed to Oxford University Press who are responsible for producing the published volumes.
Although the whole process is time-consuming it needs to be noted that for volume 2 there were nearly 62,000 name records which actually comprised some 2 million individual data items that needed to be correct. This was calculated by totalling the columns of the tables in the database, but ignoring NULL values. On the third loading run it was found that there were only 36 errors between the original data files and the unloaded database ones. 22 of these were faults in the output program, most of which were easily rectified once the problems were thoroughly investigated. The rest (14) were due to mistakes in the original data or due to missing information in the authorised checking lists rather than the actual data. The reason that these latter were not picked up by the checking programs was simply that the format of the book abbreviations could legitimately be interpreted correctly under the rules known to the program.
Given this number of errors it is possible to calculate that there was
an error rate of less than 0.002%, which in real terms is probably as
close to perfection as one can expect to get. However, one final note
of caution must be sounded. The programs can only check for format and
accuracy against information they know about. They cannot check the accuracy
of the information against the original material, so that if the reference
data includes incorrect page numbers they will remain so - the program
only knows that there should be a page number, not what it should be.
Nonetheless, a significant addition was necessary to the information in
Volume 2 and so it was decided to modify the original data files and re-load
once again, rather than modify the data in the database. The final loading
produced only a handful of errors, mostly of content. It is hoped that
the error rate for the publication of the remaining volumes will also
© Lexicon of Greek Personal Names. All rights reserved.