As part of my current project, I am collating some South Indian manuscripts. So far, I have been collating a recent Telegu transcript on paper and a Grantha one on palm leaf.

1. At the end of the training phase, I was able to collate one folio per day of the former (written in a modern notebook with 38 lines per page). This means that I will be able to collate it in full in 66 days, almost three working months. Supposing that I just register variants in a file which has already the text I am editing, I will probably save one third of the time, so let me settle for two months but the last pages of the manuscript contain an unpublished text, so that these needed to be collated anyway.

2. I just measured the time I need for the Grantha palm leaf manuscript after the training phase: 90 minutes for each side of a folio. Since the manuscript has 58 folios, this means I will need 174 hours to collate it, which means 43,5 working days (I can only collate for about 4–5 hours a day, since I cannot focus for longer than one hour on collating and I need to do something else in between), which means little more than two months. If I forget about a separate collation and just insert variants, I will probably need less, perhaps one month and ten days.

3. Next I will collate a damaged palm-leaf manuscript in Grantha containing 35 folios, which will take me little more than one month or two and a half weeks.

4. Then, a further palm-leaf manuscript in Grantha counting 153 folios, which amounts to little less than six months or four months.

5. Then, a last palm-leaf manuscript in Grantha counting 22 folios, which amounts to less than one month or two weeks.

6.–7. Last, I have a two transcripts on paper, written in Grantha in a modern notebooks, summing up to 638 pages. This risks to mean that I will have to invest 4 years on them (!). They contain an unpublished text and the collation cannot be avoided in their case.

This being said, collating in full is better than registering variants (since the latter process inclines one to read what one has in the model instead of reading the manuscript afresh) and preparing critical editions is better than accepting published texts uncritically. Still, it is extremely time-demanding (unless one enjoys collating and does it as her hobby). How important must be the text in order for a scholar to engage in a critical edition? How flawed the edition, in case of published texts? How important must be the text in order to engage in the collation of several manuscripts of an unpublished work?

Part of the problem lies also in the fact that some answers are only found while working on the manuscript(s) and the edition(s), so that an a priori answer is impossible. Thus, I test each manuscript by:

1. collating some folios at the beginning, middle and end
2. collating in any case the maṅgala and the colophon(s)
3. preparing a (keyword) description of the manuscript
4. comparing it with further manuscripts in order to detect possible transcripts, which can then be left out
5. comparing it with the extant editions in order to check whether they have already been used (which makes the possiblity of adding something significant through their collation dependent on the quality of the edition)

What are your strategies? When do you decide that collating is worthwile?