This entry started as an offshoot of another idea presented in: Clay-Pidgin: A TCP based message queuing for Tcl/Tk.

In this post, I will explain the concept of Provenance, and why sometimes the links in a database are more complex than the items tracked in the database.

I ran across the phenomenon when putting together the field researcher's database for the Franklin Tecentenary Project. The principle concern on that project was that connections between artifacts and people had a lot of properties that a monolithic filing system would never be able to contain properly. Critical pieces of information, such as dates, may not be known, or may only be approximate. All we could tell, for certain, with every case is that there as A) and artifact (or monograph, or art work,) and B) that people (or families, or organizations) either owned, stored, curated, displayed, or repaired the items tracked in A. Also, some of these relationships overlapped.

Take for instance, a lawn chair owned by Benjamin Franklin. Between all of the houses he owned, and the parties he held, Benjamin Franklin acquired a lot of chairs. Being the 18th century, nobody make "cheap" chairs like we would find today. The Industrial Revolution hadn't happened yet. All chairs were hand made, and you were primarily paying for the craftsmen's time and the material.

In this case, let us assume our quasi-fictional chair was commissioned by Benjamin Franklin's wife, and produced by a local merchant in Philadelphia. We need to record this event, and the names of both of these individuals. And the fact that one of the individuals, herself, had a link to Benjamin Franklin by virtue of marriage. As there was probably a several month gap between the order and the delivery, what date do we record? The order? Well, no. The chair didn't technically exist. However, while we have an exact date for the order, perhaps this chair was part of a large order, and we can't nail down the exact delivery date as they arrived in drips, drabs, and whatever fit in a wagon. And in fact, there may be a few dozen such chairs that all originated from the same order!

The Franklins used this chair for years, and over the course of its life, it was probably sent out for repair work. The details are probably lost to history because, honestly, who tracks receipts. But, if by some miracle, we have a few scraps of paper over the decades that this chair was in the service of the Franklins that includes an attribution of ownership, a date of the event, and either a description unique to that artifact or better a serial number, we want to record that, and be able to link it back to the artifact.

Part of the problem is that while Benjamin Franklin was somewhat well known during his life, his role as a founding father of our country was mainly towards the end. Nobody would have been keeping records on things before he was famous, so what records we have during his life are largely happy accidents from things people never threw away.

After Benjamin Passed, his estate gave this chair, and others like it, away as gifts. We may gave a date. Ish. But most likely we have a family story about how Grandfather knew one of Franklin's children, and one time at a party he snagged this as some swag. Ideally we know the name of Grandfather. And which of Franklin's children. And the date. But as we are dealing with history, sometimes even nailing down Grandfather is hard.

The chair sat in someone's salon for a while as a conversation starter. But over the years was starting to show some where. So they sent it out and had it refurbished, as was the style in the 19th century. And during that refurbishment, they had an elegant "Benjamin Franklin Sat Here" message cross stitched into the new seat cover. And as the rest of the decor in the house now had a maple finish, they had the old mahogany varnish removed and stained the chair to match. (Yes, Yes, I know there are antique fans who are utterly appalled, but I'm just reporting the facts.) This was, as best as we can tell, in the 1840s some time. Because it was in Auntie Ellen's living room at that point. (And we now need to relate any details of Auntie Ellen, the craftsman who did the refurbishment, and provide any receipts that may survive.)

And when Auntie Ellen died (as was the style) she gave the chair to the current Owner's Great-Great grandfather. Who kept it in an attic for safe keeping, and current owner only knew of it through family legend, until they had to clean out the attic recently.

The chair itself is probably a few termites holding hands at this point, but the fact it has the story attached is what makes it suitable for display in a museum.

To be clear, the above was a fictionalized account based on a composite of the sorts of data we had to work with dealing with artifacts that were 300+ years old. These problems in data integrity carry on to this day.

When the new Udvar Hazy center opened, they wanted to place a Vought OS2U Kingfisher on display. But there was a problem. There was an odd gap in the aircraft's history. The plane started it's service attached to the USS Indiana. The plane ended it's service attached to the Indiana. But historians kept finding references to the aircraft having been attached to a completely different unit for part of the war.

It also, just so happened that my Grandfather's signature was all over the logs for that particular aircraft. He was the radioman on many of its missions. So at some point my Grandfather got a call, seemingly out of the blue, from the Smithsonian. The researcher was asking for his help fill in a seemingly contradictory gaps in the aircraft's history.

There was a period during the war that the plane seemed to be in two places at once. And the curators were concerned that perhaps they might have ended up with a different plane than the one the records otherwise tell of. Fortunately my Grandfather was there for the events, and could explain what had happened.

As best as I can remember him telling the story, the USS Indiana and the USS Washington collided in 1943. The incident left both ships laid up in drydock for repairs. The aircraft needed some extensive maintenance, and were temporarily dispatched to a float plane tender for that work, and then rejoined the Indiana when it resumed operations.