Here are some ideas for when undertaking a project like this in the future.
Be familiar with the documents, preferably read through the entire document and place them into the logical structure which they will appear in. Group duplicate or similar copies of the same speech and compare them carefully--if you are completely sure they are identical discard one, otherwise keep the documents together to create the "critical text".
Once you know the order in which the documents will appear in the final form they will appear in the final document (for example, grouped chronologically, or alphabetically by location) scan them electronically once, and label them in a consistent manner, for example 010, 020, 030, (notice that each number is 10 apart, to allow for flexibility for if another document shows up and needs to be inserted between two document).
This was a major error on my behalf. I allowed the typists to submit their templates according to whatever way they wanted, and it created a lot of work to reformat so many documents into a consistent format. Especially because some people chose to "spacebar" out tabs, and others used tab--whereas in the final template a style is applied which render both useless.
Try to make this template such that if you stitch them all together they automatically number themselves correctly, etc.
OCR (Optional Character Recognition) software is improving constantly, there are some excellent options on the market, such as ABBYY's Fine Reader and others. The field is improving too, and there are options to now spend time with the machine and help to "train" it to read various passages, such as "Baha'i" which the machine ultimately rendered incorrectly.
For this project, OCR was not possible, owing to the quality of the transcripts and high proliferation of handwritten changes, however, this may not be true for your project, and even 5 hours spent training a machine will save hours and hours in the long run.
I insisted to the typists that because they had various levels of editing ability that they should reproduce exactly what appears on the manuscripts--which ultimately created more work for the editors who ultimately also had equally various conceptions of editing.
If I were to do it again, I would suggest some basic directions where changes can be made. For example: if it is clearly a spelling mistake feel free to correct it, if it is unclear what it says in the manuscript take your best guess and enclose it in [ ]s--if you want someone to review it afterwards highlight it in yellow too, use the following template and save your finished document with the same name as the scanned copy of the text.
Provide the typists with soft copies of the material, or if you provide hard copies, make these from print-outs of the scans.
This is difficult, because you have to make decisions often which only appear in one document. A safe suggestion is that the editors should have a very light hand, should not feel that they have to correct capitalisation-type issues, and stress that all major edits should be in [ ]s. Make a decision as to how you will treat transliteration, e.g.: leave it as it is, include the correct version in [ ]s or simply change it as I have.
Establish a good method whereby changes to the final version can be implemented quickly and easily.
And back-up your back-ups, for me a whole period of scanning was lost because of a very rare problem with a CD.
Copyright, Omeed Rameshni, 2009