Sunday, 18 July 2021

Chapbook Illustration and the History of Dr. Faustus

Thanks to Giles Bergel et al., bibliographers of the (very) long eighteenth-century have a valuable new widget for image-matching woodblocks. The widget was developed by Bergel et al. to search of woodblock illustrations in Scottish chapbooks held in the National Library of Scotland.

Importantly, said widget—"NLS Chapbooks"—has an external search function, allowing you to "Search using your own image" (see here) in much the same way you can conduct image searches using Google Images (here) etc.

As I have said recently (here), that the lack of a "Search using your own image" function, is the largest (remaining) limit on the utility of Hazel Wilkinson's otherwise outstanding "Compositor" (aka Fleuron 2.0).

Knowing this—and my interest in this subject generally—David Levy kindly sent me the details of the "NLS Chapbooks" search engine and a link to a very informative video of Bergel's NLS talk "Exploring Chapbooks Printed in Scotland with Machine Vision" (here).

* * * * *


Conveniently, I had the perfect candidate to test the "NLS Chapbooks" search function: the image above that appears on a chapbook History of Dr. Faustus, which I blogged about in 2010 (here). The result of my search was extremely gratifying. As you can see, it matched the image to seven items (listed below)—one of them from my copy of the History of Dr. Faustus—before the algorithm failed.


Something that should be immediately obvious, even in the small images above, is that the block has been intentionally altered—or "diminished" as Bergel explains in his talk (starting at 35.00). A halberd, a late eighteenth century version of the pole-axe-spear-weapon, is present in five of the seven images, but is missing from the two issues of the History of Dr. Faustus.

A closer examination of the seven images reveals two much more subtle changes: a very small chip in the hat (which seems to be some sort of French military bicorn hat), that becomes progressively less small, and a very small crack adjacent to that chip, which becomes progressively larger.



Above, we see the four distinct states or forms of this block represented by the NLS images, with the changes mentioned: [1] undamaged; and [2] tiny chip to hat (1st image below); [3] small chip and second tiny crack; and [4] large chip and small crack (2nd image below). The last of these corresponds with the block's de-halberdising.

Looking just at the just the top of the hat, this progression is a bit clearer—despite the pixilation at the magnification necessary.


* * * * *

The pixilation is a clear indication of the path to further progress. The excellent work by Bergel and his team may not be reproducible by Wilkinson (using images from ECCO), because the images she is working with are at a lower resolution that the NLS images. (NLS Chapbooks download at approximately 350dpi, ECCO images at 72dpi.)

Like decent OCR, matching user-supplied images requires more detailed images to reduce the number of false positives. This suggests the utility of ECCO upgrading at least a proportion of their scans with fresh photography, to improve OCR (which they now supply) as well as this sort of image-searching functionality.

But even when working with better images, such as those in the NLS Chapbooks series, resolution sets limits in identifying the sort of progressive damage to blocks seen above. This, in turn, suggests that the optimal image resolution is probably 600dpi—which has long been the digital archive standard.

* * * * *

One final point to make about the items listed below, with NLS cataloguing links, is that the suggested date of publication for these items at NLS will have to be re-considered. Judging from the ornament damage, the earliest item (Three Scotch songs) cannot be "1850–1860?" if the History of Dr. Faustus is "1840–1850?" etc.

Also, if the History of Dr. Faustus is the final form of this block, then it should also be clear that the person depicted in this block is neither Faustus not Mephistopheles—nor anyone else from the History of Dr. Faustus for that matter. As Edward J. Cowan and Mike Paterson explain:

"It was often felt necessary to ornament the front cover with a picture , and a woodcut usually served this purpose—even if it had been used several times before, was fairly crudely executed and made only an indirect allusion (if any at all) to the content." (Folk in Print: Scotland's Chapbook Heritage, 1750-1850 (2007), 13).

* * * * *

State 1: spear; no chip
Three Scotch songs: Donald Caird. Bundle and go. The Haughs of Crumdel (Glasgow: Printed for the booksellers, [1850–1860?]) [L.C.2845(30)] NLA catalogue entry.

State 2: spear; tiny chip
John Falkirk's cariches: to which is added Tam Merrilees; a capital story (Glasgow: Printed for the booksellers, [1840-1850?]), E [L.C.2852.C(10)] ¶ issue with number NLA catalogue entry; E [L.C.2848(1)]. ¶ issue without number NLA catalogue entry.

State 3: spear; small chip, 2nd tiny chip
Four popular songs: viz. Glasgow fair; Oh what a parish. A beauty I did grow; and The adventures of a shilling (Glasgow: Printed for the booksellers, [1850–1860?]), E [L.C.2845(32)] NLA catalogue entry.

The Haughs of Crumdel: to which is added, It fell upon the Martinmas time. Wilt thou go my bonny lassie? (Glasgow: Printed for the booksellers, [1850–1860?]), E [L.C.2845(5)] NLA catalogue entry.

State 4: no spear; large chip, 2nd small chip
History of Dr. Faustus: shewing his wicked life and horrid death, and how he sold himself to the devil, to have power for 24 years, to do what he pleased, also many strange things done by him with the assistance of Mephostophiles. With an account how the devil came for him at the end of 24 years, and tore him to pieces (Glasgow: Printed for the booksellers, [1840–1850?]), [L.C.2852.E(24)] NLA catalogue entry; L.C.2847(6)] ¶ issue without number NLA catalogue entry.

[UPDATED 2021.07.19]