Friday No. 5, November 2nd, 2018
Last friday I decided that my data model needed a bit more testing than possible by 31 out of 68 JSTOR entries retrieved by searching for “Relandus”. This week therefore was devoted to entering data – and having done that, I can definitely say at least one thing: I need a break. I hope I’ll make to the archive again next week. But before I’m off, are there any other results?
From 31 to 82
I projected to complete my first package of 68 JSTOR hits retrieved by “Relandus”, four which actually pertained to something else, so that made 64. And to see if this is representative data set I set up a search with the same parameters for “Braunius” to get articles related to Johannes Braun, which returned 67 hits. Looks like a good match, so I had a go at it. Deducing anything unrelated to entering data I spent roughly 35 hours this week on processing these results, which took me through all 33 remaining Reland hits and through the first 25 Braun hits (seven of which were related to something else, so 18 remained). Adding the 18 hours I did last week on 31 items from the Reland set, this makes 82 retrieved JSTOR items processed in 53 hours. On average this means that entering one JSTOR items took me about 40 minutes. This should be a bit disappointing as I only needed 30 minutes each for the first 31 and had expected the average time needed to sink rather than to rise. But to put this into the right perspective, the first 31 items netted me 93 references in total, or three per item. From the 51 items I processed this week however I made 343 references altogether – or 6.7 per item, more than twice as much as in the first part of the set. So I am not entirely unhappy with spending 30% more time per item to get more than 200% more results. In principle, at least…
Numbers, numbers, numbers
But first of all let’s take the counting one step further and see how working through this set has played out in numbers of database objects. In the course of working through these 82 JSTOR entries I added to the database:
Persons: 229; Publications: 169; Publishing Houses: 89; Institutions: 45; Web Sources: 13; and Families: 7.
In total: 552 new objects, or about 6.7 per item (yes, again…). And to put these numbers in perspective also, this meant a rise in each category by:
- Persons: 28%
- Publications: 30%
- Publishing Houses: 39%
- Institutions: 31%
- Web Sources: 19%
- Families: 17,5%.
Or, to put it simply, my database has just grown by more than a quarter in eight days. Sounds nice. But do these data tell me anything new?
Not that much, I must confess, because I haven’t yet analysed them very thoroughly. But what is directly visible is, first of all, that from the point of the references the Braun set and the Reland set are quite distinct and share only a few interconnections, which is interesting.
And while the Reland set contains more direct references to Reland himself (he’s the big red dot in the middle of the larger wheel), the Braun set contains those publications with more references at all (the two big green dots within the smaller wheel). Of course the comparison is a bit unfair because the Braun set is much smaller and not yet completely processed. But it points in the direction of disciplinary foci as the main cause for such divergences.
… and intermittent references…
And, if I may come back to my concept of a pattern of intermittent referencing as a visible marker for being structurally forgotten, so far both sets dovetail with this quite nicely.
Although an aggregated chronological visualization makes the impression of a continuous referencing pattern with ups and downs, in a more detailed view the three main clusters (1878-1908; 1918-1948; 1968-2008) break down into singular spikes and large plateaus.
… and the catch
Of course there is a catch. If this would be all there is to it, were would the challenge be? Well, the catch is that this is just the tip of the iceberg. If I perform the same search for “Reland” instead of “Relandus”, JSTOR returns 596 results instead of 68.
And querying just for “Braun” is plain nonsense. Even with “Johannes Braun” it’s still tricky to sort out mishits. The same goes for “Thomas Gale”; “Eusebè Renaudot” works better, but returns a corresponding number of hits. Taken together these searches amount to about 700 or 800 real hits, roundabout ten times as much as I have now processed in eight days. And even if I chose to spent 80 days only working through those JSTOR results, this is only one platform of many to be queried. Which means am back from where I started. It will not work out like this. I have to find another way to deal with this. When I’m back from the archives.