Mistakes I Made While Organising a Collection of Rare Marathi Manuscripts

When three donated boxes arrived at the back room of the library, I had opened a spreadsheet before I even properly figured out what was inside them. It was a collection of Marathi manuscripts sent from a family on the outskirts of Pune. Forty-seven items in total. The condition of the paper was uneven, and several pieces had clearly visible wormhole marks. I told my supervisor I would complete the full catalogue within two weeks. That confidence was my first mistake.

The Price of Oversimplifying Provenance

I summarised what the donor's granddaughter explained over the phone on a single sheet of notes. "Collected by grandfather in the 1940s, some estimated to date from the late 19th century." I copied and pasted this one line into the provenance field for all forty-seven items. Even though the texture of the paper, the density of the ink, and the handwriting differed from manuscript to manuscript, I lumped them all together as items from the same era, acquired through the same channel.

Later, a conservation specialist examined three of the items and told me that one of them used an entirely different type of paper from the rest. It was handmade paper based on cotton fibre, while most of the others were modern paper mixed with bamboo pulp. Because I had lumped the provenance together, this difference went unrecorded. If that single item was an older manuscript that had come in through a different channel, that information could have been of real scholarly significance. The time I saved was three hours at most, but it took five times that to fix.

If I were doing it today, I would record the donor interview, describe the physical characteristics of each manuscript individually, and only then enter the provenance information. It is slower, but once you record it properly, you never have to go back.

Assigning Condition Grades by Eye Alone

The condition grading system used by the library had four levels, from A to D. I would open a manuscript and make a quick judgement like "edge tearing present, Grade B." The problem was that a considerable number of the Marathi manuscripts had been stored folded. Even if the visible outer surface looked clean, there were cases where mould traces were present on the inner folds or the ink had bled, leaving the text smudged.

One item I recorded as Grade B and placed on the browsable shelf, but when a reader unfolded it, they reported that the ink damage on the inner surface was severe. That manuscript had to be immediately transferred to the conservation treatment waiting list. If I had unfolded and checked every surface from the start, I could have taken protective measures before any reader's hands touched it.

It would have taken several more days to unfold and check all forty-seven items one surface at a time, but that is not time you can trade against the lifespan of a manuscript. Now, when assessing condition, I describe at least three surfaces separately (front, inner fold, and back) and take photographs to attach to the digital record.

The Idea of Filling In Metadata Later

The hardest habit to break was the attitude of "just enter the basic information for now and add the details later." I would input only the title, estimated date, and condition grade, leaving fields like subject classification, language variant notation, and presence of illustrations blank. My logic at the time went like this: you need the full catalogue first to set priorities, so the blanks can be filled in later.

But "later" never came. Other work piled up, and the empty fields triggered no warnings in the system. Three months on, when a researcher searched for 18th-century Marathi religious texts, our collection contained a relevant manuscript, but because the subject classification was empty, it did not appear in the search results. That researcher used a holding from another institution, and our manuscript might as well not have existed.

From then on, if there is a field I cannot fill at the time of entry, instead of writing "unconfirmed," I add that manuscript to a separate review list. Every Friday afternoon, I pull two or three items from that list and fill in the detailed information. It is not perfect, but it has definitely reduced the number of blanks left unattended.

What the Slow Method Left Behind

Finishing the work on this collection ended up taking two and a half months. Five times the two weeks I had initially expected. In the process, it came to light that one of the manuscripts was a rare document written in Modi script, and it took ten days just to contact the donor's family again to confirm additional context. If I had rushed to finish, that document would have been buried as "Marathi prose, date unknown."

What I have learned working at a library in Mumbai is that manuscripts do not tell their stories to those who rush. The next time a donated collection arrives, I will take out my notebook before I open the box.