Up to now, old-spelling IML texts have been processed by MB using his own tools to generate a version which forms a useful basis for a modern-spelling version, to give back to the editor. I was asked to replicate that process so that JJ can provide these versions for editors once their old-spelling version is finished. I've implemented a transformation scenario in Oxygen which uses a combination of XSLT (working on unparsed-text) and ant with JavaScript scripting to do the job. Test output is now ready for checking, and if it all works, I'll set up the scenario for JJ in her Oxygen (or perhaps create a special project file for the purpose). The code is in trunk/sgml/utilities.