Clarify the status of GEN (Unknown) sources

[martindholmes]

One item in the sources list consists of the id "GEN" and the gloss "(Unknown)". There are 1028 links to this item, in 989 different trader records. Here is a collection of example records that have three or more such links:

https://hcmc.uvic.ca/project/bbti/collection.html?collTitle=Records%20with%20many%20links%20to%20GEN%20source&records=124036;122074;50459;95983;65347;122107

In many of these cases, a clear source is given in a note, so the link to "Unknown" is, on the face of it, erroneous. It seems to me that the pattern of assigning this source to a record, then adding a note with a true source in it, is mostly undertaken when the source is a text which is unlikely to be generally cited in other records, and therefore perhaps does not merit inclusion in the formal list of sources on the sources page.

However, the resulting record appears a little confusing, so I wonder if we might address these cases by perhaps adding another pseudo-source whose gloss/description better explains the situation, and reserving "Unknown" for cases where the source genuinely is unknown. Also, there may be some cases where a source does merit adding to the main sources list, and could then be re-linked in the conventional way.

[DavidJamesShaw]

Yes, these 'GEN (unknown)' notes look quite confusing.
If it is possible to make some improvement, I can't see any objection.

[martindholmes]

@DavidJamesShaw This is one of the many cases we'll be dealing with where we ought to keep our decisions and changes in line with Allison's, so perhaps we should agree on a specific approach here.

[martindholmes]

There are three distinct cases:

1. The source pointing to GEN has no other info, or perhaps has a note consisting only of "?". I believe these are genuine cases where the source was/is unknown.
2. The source pointing to GEN has some additional information in the ExtraInfo column, such as a British Library reference number. Presumably these cases are not unknown at all, but for some reason were not fleshed out into complete entries or complete notes.
3. The source pointing to GEN has a fully-fleshed-out note (and possibly also some ExtraInfo), meaning that the source is actually known, but presumably was not worth including as an entry in the main sources list.

My suggestion would be:

• Create a new source list entry with the caption "Specialized source", and switch all cases where there is extra info or a note of any length (cases 2 and 3 above) to point to this source instead of "GEN". In the web page view of these entries, we could add a caption such as "See information below".
• Leave the remaining cases, which presumably are genuine cases where the source is unknown, but expand the explanation "Unknown" to specify what it actually means (whatever that is -- there was a source but it got lost somehow, or the information is in someone's memory but they don't know where it came from, or what?).

[allisonmuri]

When I volunteered to do some data entry in 2007, there were some glitches with the forms. I wrote to John Hinks to say that there was a problem with one of them, where trying to add data from a dropdown was generating an error message. He said he'd talk to IT but it seems support was not all that readily available. All of this is a preface to a hunch.

GEN probably originally was meant to provide a general category where the source was indeed unknown. My hunch is, possibly, that the only access people who were doing the data entry had to make changes to the database data was via these glitchy forms and they couldn't make direct edits to the database, so couldn't simply add "The interesting history of Mother Goose (R. Carr, printer, 101, Houndsditch, London, (n.d. 184-?)" to the sourceshtml table in the source column where it would make sense for it to be added (this is the Carr example). So my guess is that they did the cludgy workaround, and put the known source into the extra_info column of the tbltraderid_sourcecode table.

The extra_info column is normally used for truly extra information about a particular source, such as in the William Birkehead and Henry Archer entries, where the source is the same, Yorkshire Archaeological Society Record Series (source code JAL/YASRS), and the extra info is the particular volume number where the reference to the individual booktrader appears.

For the most part, each row in the source column of the sourceshtml table is a single text with its own unique source code. Eventually I expect it would make the most sense to create new entries for those like "The interesting history of Mother Goose" in the correct column of the correct table.

I don't think the source codes such as GEN should appear on the booktrader pages (they didn't in the original BBTI); I contemplated not showing them at all on the Source List either, but that would have meant mucking around with the display that they originally had (the order would either be oddly un-alphabetical if ordered by a source code name that didn't show, or I'd have to re-order the list), so I just reproduced their original display. I suspect the only reason that people might have wanted to see those codes in a handy list was to help with making data entry a bit easier.

[martindholmes]

@allisonmuri Thanks for this clarification! It does make a lot of sense. There's obviously a lot of remedial work to be done in this area of the dataset. I wonder how easy it would be to de-duplicate the sources listed in notes?

Some would be easy; there are for example 8 instances of "Salter, H.E., Oxford City Properties. Oxford Historical Society, 1926.", differing only in their following page numbers.

Others are more mysterious; there are four instances which say simply "Mada", which it probably not the Malay word for "not yet"; perhaps it's intended to be a pointer to an entry for "F. Madan, 'The Daily ledger of John Dorne, 1520' in C.R.L Fletcher (ed.), Collectanea: first series, Oxford Historical Society, Oxford, 1885, 71-177.", which also appears once.

There are 86 instances of "Guildhall Library MS", mostly pointing at MS # 11937 but also # 14386.

"Stokes, H.P., Cambridge Stationers, Printers, Bookbinders, &c. (Cambridge: Bowes & Bowes, 1919)" appears 85 times with minor variations.

Such semi-consistent patterns could easily be gathered up and converted to source items once we know what the canonical data source is going to be and where it'll be stored.