Who printed Piquet for Francis Cogan? Thank you Compositor!

(udpated 3/13 to link to Patrick Spedding's post, with his discussion of Compositor)

Last week, I watched Joseph Hone present a paper 'Secrets, Lies, and Title Pages' (now available on YouTube) sponsored by ODSECS, the Open Digital Seminar in Eighteenth-Century Studies. Joseph discussed how 18c printers "corrupted title pages with false names, dates, and places...to disguise the origins of dangerous  books or piracies." I recommend the talk highly. 

In general, we don't know who printed a particular book in the 18c. The imprint, if honest, tends to identify the publisher and only occasionally the printer. Several times, Joseph made reference to the Compositor database of 18c printers' ornaments which he used to unmask printers who would otherwise have stayed hidden. I knew that Compositor was an upgraded version of Fleuron, a site I had used frequently. In my online Hoyle bibliography I had links to Fleuron which no longer worked in Compositor. It had been on my list to update the links and Joseph's talk prompted me to do so. Done!

I also explored Compositor and was blown away by a new feature: "image search", with a tutorial on their blog. It allows you to take an ornament in one book and find matches in others. Sometimes, those matches will be in books that identify the printer, suggesting a printer for the original book. Before giving an example, here is some background:

• Printers ornaments are decorative elements, generally woodblocks, used in books through the late 18c. For a charming example, see the squirrely headpiece here. 

• The source for Compositor is ECCO, Eighteenth Century Collections Online, a subscription database of nearly 200,000 18c books that can be accessed through most university libraries. ECCO began as microfilm and was digitized from the film. For a great history of ECCO, see Steven H. Gregg's paper "Old Books and Digital Publishing: Eighteenth-Century Collections Online" available here. The path from film to digital means that the reproductions are not always of pristine quality. 

• The Compositor/Fleuron team must have done an immense amount of image processing that I can imagine only vaguely. They extracted the ornaments from full pages, developed ornament metadata, and, most magically, allowed visual search. Well done!
  

Okay, enough talk. Let's figure out who printed Piquet for Francis Cogan. My description of the book is here. The imprint, "Printed for F. Cogan at the Middle-Temple-Gate", is silent as to the printer. Scroll down to the contents where it says ‘[headpiece] | SOME | Rules and Observations | FOR | Playing well at CHESS. | [...]’. Click on the link to see again the headpiece with squirrels and then click on "Load Ornament in Visual Search". 

Now for the part requiring some dexterity. As described in the tutorial, you can use the right mouse button to select a rectangular area in the ornament. It will highlight red as you drag, and turn yellow when you are done.  

Selecting an Ornament

When I clicked "search", I found 103 matches: 

Matching an Ornament

You can click on a match and the original image on the left and the match on the right. If you click the middle image, it will toggle between the two and you can determine whether they were made from the same woodblock. Many things can account for differences even when the block is the same: Woodblocks become worn from use. Any given impression can use more or less ink. The microfilm and digitization can introduce artifacts. Despite the differences, I'm awestruck by how well this works! Truly, this be miraculous!!

Now for the part I found a bit clunky. You can click on the filename of the rightmost image to go to a page like this and then click on the link "This ornament was extracted from this book". In this example, you see the first of ten volumes of Moliere's works with the suggestive imprint "printed by and for John Watts at the Printing-Office in Wild-Court near Lincoln's-Inn Fields". One match does not a printer identify, so you'd want to look at more ornaments and more matches. That entails a lot of clicking. And a lot of keeping track of what you're seeing. Well, I wondered, couldn't I automate that?

Did you notice the little button that let you export the search results as a .csv (comma separated variable) file? Well, I saved the 103 matches to a file and dragged out some rusty Python skills to read the .csv file, visit the 103 matches, visit the book from which the ornament was extracted, extract the imprints, and print them out. Ninety minutes of coding; sixty lines of code. It took longer to write this blog post. If you run it for the 103 matches of the squirrel ornament, the first eight results are:

filename:  105540010000600_1
ornament ID:  1171998
ESTC:  T048220
publisher:  printed for F. Cogan at the Middle-Temple-Gate

filename:  005770040202750_0
ornament ID:  944832
ESTC:  T052789
publisher:  printed for H. Lintot, J. and R. Tonson and S. Draper

filename:  041920010801090_0
ornament ID:  634261
ESTC:  T064098
publisher:  printed for John Watts at the Printing-Office in Wild-Court near Lincoln's-Inn Fields

filename:  025940030001100_0
ornament ID:  1028312
ESTC:  T064113
publisher:  printed for John Watts at the Printing-Office in Wild-Court near Lincoln's-Inn Fields

filename:  086630010201970_0
ornament ID:  166790
ESTC:  T089176
publisher:  printed for J. and R. Tonson and S. Draper in the Strand

filename:  025940040000700_0
ornament ID:  838797
ESTC:  T064114
publisher:  printed for John Watts at the Printing-Office in Wild-Court near Lincoln's-Inn Fields

filename:  094430010500090_1
ornament ID:  706629
ESTC:  T064441
publisher:  printed by and for John Watts at the Printing-Office in Wild-Court near Lincoln's-Inn Fields

filename:  015270010000960_0
ornament ID:  763509
ESTC:  T063293
publisher:  printed for Jacob Tonson in the Strand 

It would have been easy and useful to extract the title and date. And to replace the ' with an apostrophe. But that would have taken more than 90 minutes. Had I done so, you would have seen the first item listed is the source book Piquet. Only one of the eight imprints identifies the printer: "printed by and for John Watts...". Of the 103 entries, 31 of them include "printed by" and in all cases, the printer is John Watts. I've visually inspected a good number of the ornament matches and similarly checked other ornaments from Piquet. I'm completely confident that I have identified the printer. 

There are some caveats in working with ornaments to identify printers. The printing of a book may be shared by more than one printer. A printer may loan out his ornaments. You have to be careful about when a printer died--another printer may have inherited the ornaments. My sense is that these caveats are mostly (repeat mostly) theoretical, but you should be aware of them. 

In fact I had done a lot of pre-Compositor ornament searching and had already identified Watts as the printer of Piquet. But with this new tool, I have identified printers for some of the Dublin Hoyles and for some non-Hoyles in my collection. 

Thank you Compositor! And thank you Joseph for the nudge!

For Patrick Spedding's take on Compositor, see his blog essay here.