A guest post by William Davis
Thank you to everyone who left a guess on this month’s crocodile mystery! Everyone got a piece of it, but none the whole. It takes a stalwart person to identify some of the many quotes tossed off by the lead character of “Cupid’s Whirligig!” The Black Spot is truer than you would think: it is indeed a mark of death. Finally, it is true that this is a digital image of microfilm, but it’s not a burnt through place on the film! It’s much more intentional than that. This image shows an actual hole punched through the microfilm. Why did they do that? Well, when the microfilm camera operator accidentally shot two images of an opening or made some other error, they could “delete” one by making a hole in it with a hole punch, instead of splicing it out. The user would know to skip over the punched image.
If you want to know more about this (and other oddities found in microfilm), read on!
If you are one of those special people who wakes up each morning, fires up the computer, and goes to the Folger’s digital image database (Luna, at present) to see what’s new, you will already know that The Folger Microfilm Image Collection has slowly and quietly oozed into existence recently. It doesn’t have many titles in it now, but we plan to change that over time.
When a call number is uploaded into this collection, it consists of digital copies of all the images in the original microfilm. The key word here is “all”. Sometimes it includes more, sometimes less, and sometimes different things, than one expects. We have already received questions about what exactly is going on with the omissions, duplications, and weird images that are sprinkled through this new digital image collection.
Familiarity with a microfilm of a rare book is becoming more uncommon as time passes and fewer and fewer people have used actual microfilm to explore the early modern period. Most odd things visible on digital images of our microfilms happened as a result of microfilming an item or using the resulting microfilm, while some are due to the largely automated process of digitizing the microfilm. This post is an effort to explain at least some of what you are looking at when you aren’t looking at what you expect to see in the new Folger Microfilm Image Collection.
Microfilming at the Folger
Microfilming at the Folger changed significantly through the decades before we stopped microfilming in the early 2000s. We began microfilming near the end of the Truman administration (around 1950), and there have only been industry standards for how to microfilm since the 1970s (Nixon, Ford, Carter). This means that the Folger now has a lot of microfilm created by relying largely on the customs of the institution and the intuition of the microfilmer, both of which changed over time, as did industry standards later on.
Microfilming in its heyday was a pretty low-tech operation. This was doubly true at the Folger, because we purchased a Kodak Recordak microfilm camera around 1950, and since it never broke irreparably, we never bought another. This image of a U.S. Navy Kodak camera differs from ours only in that it is set up for large flat items, such as maps or architectural drawings, while ours had a large, moveable box on the table. The box contained a flexible book cradle supported by large springs that would press the opened book up against the glass box lid for optimal imaging by the camera above. The cradle could accommodate books up to about 14 inches (35 cm.) in height. Larger items could be filmed by removing the box and building a cradle from foam and other supports.
The camera itself was adjusted by hand. It could slide up and down the central pole, and move 90 degrees side-to-side in its bracket. Sliding it down the pole would reduce the reduction ratio, while pushing the camera up would increase it. The smallest reduction ratio we used was around 8:1, producing an image that was 1/8 the size of the original, used for duodecimo books (approximately 5”x7.75”) or smaller This would enlarge the image so that it came close to filling the width of the film, making the image easier to read. The largest reduction ratio was around 24:1, used for maps or extremely oversized books that were too large for the box. This would shrink the image so it also would fit on the film. Most books were shot at 12:1, conveniently sized to 1 foot (12 inches) on the book reducing to 1 inch on the microfilm. This example works well with Imperial measurements, not so much with metric; for that you need to shoot at 10:1.
Swiveling the camera back and forth 90 degrees, between straight towards the operator (position A) and the perpendicular (position B), allowed the operator to align the film edges with the long side of the item. If the item was taller than it was wide, or so small that the opening would not fill the space between the film edges, one could ensure that the images are as large as possible by shooting it in position A:
If the opposite were true and the item’s width was greater than the height, the operator would move the camera to position B:
There were two different types of lights on the Recordak camera:
1. A set of four large shooting lights (2 on each side of the camera, with one at each corner of the item being photographed) that illuminated the entire table and were controlled by a dimmer (voltage regulator) and an on/off switch. These four dimmable lights changed the exposure (dimmer to avoid overexposing light manuscript on white paper, brighter to avoid underexposing dark paper or print bleeding through the leaf.)
The operator had to estimate the proper voltage as the paper changed through a book. This happened quite frequently with rare materials, such as when the title page darkened over time due to higher use. If the operator’s decisions were correct, and the laboratory technician developing the film made the correct timing and chemical strength choices, the result would be a readable film: one on which every image was neither overexposed (a brilliantly white image where the writing fades into the glare) nor underexposed (a dark image where letters can’t be picked out of the gloom).
2. A single viewfinder light in the camera head that illuminated the area on the table below that was captured by the camera lens. The viewfinder light turned on automatically when the dimmable shooting lights were turned off. It projected the complete view seen through the camera lens, with lines denoting the sides of the usable image, as there were problems with attempting to shoot right to the edges of the film. The viewfinder would confirm that the proper reduction ratio was chosen, and the size of the image filled up the optimal width of the microfilm for maximum readability. A dial in the camera head could lengthen or shorten the ends of the frame perpendicular to the edges of the film. This allowed significant expansion of the frame for fold-outs, or contraction for a single loose leaf, for example. Not checking this viewfinder regularly and making adjustments as necessary during filming (for instance, to accommodate fold-out leaves) could lead to skewing or to images with an edge entirely outside the frame. It could also waste film if the frame was sized larger than necessary, allowing fewer exposures in a given length of film.
What are the oddities most often found in the digital images in the Folger Microfilm Image Collection? Apart from mistakes in exposure and framing, problems due to microfilming generally involved shooting errors. Unlike a digital camera where you can immediately view and delete any bad images, a microfilm camera was unforgiving. If the operator’s concentration slipped for a moment, the only way to confirm if a mistake was made was to check the film after it was developed. On the one hand, the operator might skip one or more openings in the book. If this was discovered during checking after the film had been sent out and developed, a correction was shot. The correction was sent out, developed, and checked in turn. If it was correct, it was then spliced into the original film. Then the corrected film was sent out for duplication, and when the positive duplicate came back in the mail the order could be filled, sometimes a month late or more. Since the patron was routinely given a turn-around time of 6-8 weeks, this was a hardship. On the other hand, the operator could accidently shoot an unwanted extra image, for example of an opening, or one’s hand turning the page, etc. This was generally not considered a major problem.
A second source of oddities are changes in standards and best practices over time. This is seen most clearly in the wide variation in targets used through the years. Also, as technology improved, three different types of splices eventually were used in our microfilm collection, from glue, to tape, to ultrasonic splices.
The third source of oddities is the life of the microfilm as a service copy. It was only in the 1990s that the Folger began following standards for microfilm storage, by putting the microfilm on 100’ reels, storing the originals separately, and creating a service copy for public use. Before then, the service copy was the original microfilm. This meant that rips in the film had to be taped back together as best they could be and the film sent back to the reader, rather than having a new positive duplicate made from a pristine original film. Also, there was nothing to be done about scratches created by the film repeatedly sliding between the glass plates of a reader-printer. In the worst cases, these scratches could affect readability, when positioned exactly over a line of text.
The final source of oddities in the digital images made from the microfilm are glitches in the largely automated microfilm digitization process. There are few such problems, but they are distinctive. We contracted with an outside lab for this service and have never done it ourselves, so we are not clear on all the details, but the general process is clear. A microfilm reel is loaded into the automatic microfilm scanner. It spools the film through, essentially making a digital version of the microfilm reel. Then an autocropping program is used to automatically crop out each photograph on the film and create a separate image file for it. This autocrop feature works very well, say, perhaps 99.9% of the time. The problem: 0.1% of 700,000(+) images is still 700 images! So it is not surprising that occasionally we will mention that the autocrop did not make the right choice. Sometimes it is fixable, and we do so (while keeping a copy of the original, because we are librarians). Other times it is not, and then it must be recognized that there are some things that go irretrievably wrong when attempting a project of this size, but the benefits outweigh the drawbacks.
I. Old Conventions/Standards
One of the longest standing conventions in microfilming is that every rare item microfilmed should have initial and end targets that identify the item in question, containing the institution, the call number, and bibliographic information. In addition, occasionally a situation would arise that would require interior targets, such as a warning that a number of blank pages are not filmed, or a notification of a change in reduction ratio or orientation for a very large foldout. While these targets were well standardized by the 1990s, older Folger microfilms use a bewildering variety of different targets, and some early microfilms have no targets at all. The same variety can be found in call numbers and bibliographic information, which can change over time as an author is identified, or a year of publication is found to be incorrect on the title page, or a new call number system is introduced. The metadata (the searchable information associated with the image on a computer) for the digital images should be current, but don’t be surprised if the information on the digital image of a target for a microfilm created in 1952 bears little relation to the metadata.
An early initial target:
Miscellaneous Superseded Conventions
1. Highlighting a partial leaf:
An early method to isolate a partial leaf by inserting a piece of white paper behind the leaf. This method was not used later on. Fortuitously, the second image follows later convention, where no effort was made to highlight the partial page. The tear or cut lines across the leaf still should be visible on such images, though.
2. Background not set to black:
II. Filming Problems
2. Edge of item out of view
This is a major error. It seldom occurs in the Folger microfilm collection, because when found all the affected images would be re-shot so the reader could read everything on the page, or at least be certain nothing is written on the blank parts of the page. One example of a call number with this error is STC 837 copy 2 (see especially this image and the image above); a second call number, with many more images affected, is W.b.422.
In many images in W.b.422, the entire bottom is out of camera view, cut off along the bottom edge. While the reduction ratio on the camera was set large enough to show the entire book and perhaps the flag (image 1), it should have been higher, making the images smaller. Clues that images go out of frame here include that some images are skewed (images 2 and 3), but the outline of the book creates a trapezoid (image 2) or a five-sided polygon (image 3), when the book is definitely rectangular (image 1). Secondly, in image 2, writing is cut off in the lower left-hand corner along the bottom of the page. This is not unusual when a book is rebound, however, so if the image is not skewed in this way, there may be another cause for loss of writing.
Exactly why the film of W.b.422 was passed by the checker is unclear. It is possible that this problem went unnoticed, although one could also imagine that since the manuscript is long (about 180 openings) and most pages of W.b.422 only have manuscript on the top half, the checker decided to declare no harm, no foul. While the checker, comparing the images to the original manuscript, could see that there was no manuscript at the bottom of most pages, however, the reader looking at the microfilm now cannot confirm this. Sadly, the problem is worse than that, for clearly text is lost on some pages, and now the reader can’t be sure why. what it was, or if more was missed.
3. Duplicate images of various sorts
Multiple Exposure Duplication
Making two images of an opening, one at a high and another at a low exposure (voltage, or film density) is not an error. It was done in order to allow the reader to clearly read all of the writing in the opening. The exposure of image 1 was adjusted to allow the reader to see the light manuscript on white paper, underexposing the dark areas due to low voltage. The exposure of image 2 is much higher, to capture the dark manuscript on a dark background, overexposing the light areas due to high voltage. The detail images below make this easier to see, as both the dark writing (at the very top) and the light writing (at the bottom in the margin, to the right of the gutter) are visible in both images.
Not a Duplication at All
These two images of opening 57 are different from each other, because there is a 3×5 card with notes on has been laid in (inserted, like a bookmark) in this opening. The first image shows one side of the card, the second image shows the other. One can tell (rather fortuitously) that the card is not tipped in (glued down along one edge to leaf 56 verso, as postal letters often are) because it shifts both vertically and horizontally between the two images. Sometimes the difference on microfilm between two images can be very subtle.
“Did I shoot this opening yet?” The microfilmer couldn’t remember. Therefore, they shot it again to be sure before turning the page. Regrettably, they had indeed already shot it, and so they ended up with two images when only one was needed. Alternatively, their foot jiggled and hit the shutter release twice instead of once. We know that this is not a correction because the density and the edges of the opening are identical, which is practically impossible to accomplish when microfilming the same opening on two separate occasions.
An Unacknowledged Correction
A bad image here is corrected by the next image. The first image was shot while book box glass lid was raised (the shadow across the opening is from the frame surrounding the glass); the second image is the readable re-take. In later years, the first image would be hole punched (see below); even later, it would be removed and the film spliced (see further below).
An Early Standardized Correction
A Modern Correction Duplicate
After the discovery of a missed image in a microfilm on file, a correction was shot decades later using current procedures and standards. The deeper gutter on the correction image on the right makes it likely that the book was (re)bound before the correction was shot. The original image on the left is severely overexposed, making it appear washed out. The exposure of the correction is within standard tolerances, and shows just how overexposed the original film is. The two books are actually shot at the same reduction ratio, but the inclusion of the flag changes the size of the thumbnail image of the correction.
Here is the entire correction. The newer standards required the use of a call number flag in each image. It was also necessary to shoot a series of images in the correction, from a duplicate of the last correct image before the error, to the missing image(s), to a duplicate of the first correct image following the error. The previous and following correct images in the original film were retained, resulting in two sets of duplicate images, so the reader could tell that the microfilm is complete. In the example above, the first two images and the last two images are the same opening, and the center image is the missing image. Prior to this era, the emphasis was on supplying one image of each opening only.
The Folger has used three types of film splices over the years: glue, tape, and ultrasonic. Above is a beautiful image of a glue splice, the oldest type of splice found in the Folger collections. In order to smooth the splice so it didn’t catch on the glass plates of the microfilm reader machine, the film was shaved on the underside of one edge and on the top side on the other. Since the actual image is only on the underside of the film, in this example the image is shaved off of the left edge, making it translucent. You can thus see both edges of the splice.
A major disadvantage of glue splices was that over time, the glue would dry and become brittle. Twisting the film in this condition can crack the splice apart, scattering glue fragments. In the image one can see how the glue has deteriorated at the bottom of the image, below the book. As a result of the transient nature of these splices and the time required to make them (shaving, gluing, drying), glue splices were eventually replaced by polyester tape splices. The vertical splice here means that the book was microfilmed in camera position B, with the gutter perpendicular to the edges of the film.
Tape splices, which required a specialty tape, were an improvement, both because they lasted better than glue and because they were much easier to apply. One would use a tape splicer, which had everything you needed. You would cut each end square, place the ends of the film touching each other, tape across, flip the film, and tape the splice again on the other side.
Here is a complete correction using tape splices of another book shot in position A. The underexposed image was cut in half, and the correction inserted in between. Each tape splice is incomplete in these images, as the digital outcrop did not notice the edge of the tape that isn’t on the image, but rather is on the black film in between.
I was able to reconstruct the complete first splice, however, oriented as it would be seen on the tape splicer, with the film running right to left.
Sadly, I haven’t yet found an image of an ultrasonic splice, which were developed for polyester microfilm. Virtually all of the microfilm in this collection is acetate microfilm.
5. Pictures of Hands and Other Body Parts
Conscious Effort to Protect Item While Allowing Unobstructed View
Another not-a-mistake: Early on, it was acceptable to use hands visible in an image to safely maximize readability by holding curling edges, etc. while filming out-of-the-box. Here is a vellum document held open by two assistants, and a scrapbook, with a hand holding the left page out of the way. Essentially, seeing something and a hand was considered better than seeing nothing without a hand.
Images of Body Parts That Shouldn’t Be There
All the following are microfilmer’s mistakes.
III. Problems with Microfilm in Public Use
1. Tape splices repairing ripped microfilm
These are rather bad jobs (which are easier to see than the good ones when scanning film.) They were either made by Folger staff attempting to make the microfilm usable again for a researcher, or quick fixes by the digitizer when the film ripped and ground to a halt in the digitizing camera.
Another example. This microfilm now appears to be angled over the rip (see the right-hand side), which would not fit on a microfilm reel easily.
The angle appears to come from the two sides of the film overlapping under the tape, rather than fitted together smoothly, making the image difficult to read (see especially the first image).
2. Scratches on Microfilm
Microfilms of important and restricted items often ended up with significant scratching due to heavy use in microfilm readers. The film had to be pressed flat between two pieces of glass to make focusing possible. While the top glass did not lay directly on the bottom one, the distance between them was minimal, and scratching was easy, especially if the user was not careful to advance the film slowly.
IV: Problems Introduced (or Not Corrected) by the Digitization Process
Now is the time to comment on what we have corrected in these digitized microfilms, and why. We sent out approximately 7,700 microfilm call numbers for digitization, which produced 747,329 individual digital images. We have not corrected any problems described before this point in this post, for two reasons:
- Librarians can be very conservative when changing things, of whatever sort. Ideally, a researcher should be able to reconstruct what previous generations saw, because you never know when it will be important. So we keep records of changes. An example would be call numbers. A researcher looks at a manuscript in 1956. They note the call number: MS 49.4. The call number is later changed to V.b.57. A researcher in 1998 reads the 1956 journal article, is intrigued, and asks to see MS 49.4. Unless we have the history at hand, we can’t give them V.b.57 without a lot of flopping about, or luck, or both. Happily, in that case we can point them to V.b.57, because the older number is still in the online catalog record, and is in a note in the in-house microfilms database as well, to name only two locations. This means that for the oddities in shooting and using the original microfilm, what you saw then is what you see now, warts and all. One of the major reasons for this post is to explain to modern users what they are seeing, because familiarity with old microfilming conventions is being lost as the technology changes.
- Removing the ugly, the mistakes, and the duplications would be an incredibly time consuming process. Approving each individual image is impossible given current staffing and other projects. We only had to do the math once: 747,329 images at 5 seconds per image requires one person to do nothing but look at images, without slowing down the pace while making corrections, for 29.5 weeks. Oddly, no one volunteered, and we weren’t so hard-hearted as to strongly suggest that they do so if they knew what was good for them.
However, with respect to the major problems (that is, problems that affect readability) resulting from the digitization process, we have made an effort. We have already looked briefly at each image folder in thumbnail format and fixed all the major problems we could find that came from the digitization process. While I hope that we found them all, of course it is likely that we missed some. We regretfully left it at that, with a plea that if you find anything that must be fixed when looking through the digitized microfilm, please let us know!
Instead, we decided to focus our limited resources on linking all 7,700 call numbers to the appropriate online catalog records, to make the collection truly searchable and usable by everyone. Although that is an extremely time-consuming project, it will pay the biggest dividends over time; it is worth the trouble.
With that in mind, here are problems introduced most recently into the microfilm collection as it underwent the process of becoming a digital resource:
1. Garbage In, Garbage Out: Problems likely due to the microfilm being wound on the reel incorrectly
Over time and extensive use, some microfilms are bound to get wound on their reels incorrectly. This generally happens when a researcher either puts the microfilm on the reader incorrectly (some reels can only be put on a machine in one way, to prevent this from happening, but they are not always used). Then they have to figure out how to re-wind the film so it is readable. I hope this is the case, as I can’t figure out any other way it might happen, and it does happen. Anyway, you have four possibilities:
The question then becomes, How much gets corrected during the digitization process? Does the operator see that the film is not oriented correctly, and change it right off? Does the machine autocorrect all of the problems, or some of the problems? Is there a human checker (well, there is me, at least), and do they catch all or some of the remaining problems? As I have said above, it is quite difficult to catch every major problem in every image file. I now see why nameless software companies throw new software versions out there and let users tell them what’s wrong, rather than taking years to try to de-bug it themselves. We have tried to do better here, but make no promises.
Example of Position II:
The beginning of a microfilm with digital images numbered in reverse order. The images are readable, so somewhere along the line each one must have been spun 180 degrees. But they still read back are still backwards. This situation is the most common of what might be called the winding problems.
An Example of Position IV, or Perhaps Position III (with the images each spun 180 degrees):
I have completely corrected every example with images flipped along the vertical axis that we found, so this is a made-up example. As best I can remember, the actual files with this problem were in the correct order (they start at the beginning and finish at the end), but I can’t be certain. Unless you catch one with a large END target, they can be very difficult to identify when looking at thumbnails because almost all of them look like this:
Which is really this:
Other, Unrelated Upside Down Oddities:
This is an upside down target in an otherwise right-side up microfilm. Unlike digital images of entire books that have been found in this digitization project, this image is likely the result of the microfilmer’s inattention rather than a problem with the way the microfilm is wound on the reel. Since a microfilm reader allows one to turn an image 180 degrees, this was probably not considered worth correcting on film. Once the images are digitized, however, it is a very easy permanent fix. Since it is so easy, this is a made-up example.
The above mistake is superficially similar to the common situation below, where an extra-tall target was purposefully shot sideways in order to fit it into the image boundaries of a lower reduction ratio that is optimal for the book, which is much smaller. We have not corrected these side-ways targets throughout, as they are easier to read than an inverted target, and the digital image database allows the reader to flip an image 90 degrees in either direction if they do not wish to look funny while doing research.
2. A Problem With Universal Commands
Since almost all of the microfilm digitized in this project is original film, that means that it is negative film. Only positive duplicate copies made for public use have standard black-and-white images, such as the examples seen in this post. However, while the Folger sent positive duplicate copies to researchers to fill microfilm orders, we did not make positive duplicates of microfilms for public use until the 1990s. Therefore, if one looked at microfilms while doing research on-site at the Folger Library before that time, one would be reading negative images. This means that for this digitization project, in order to produce positive digital images, there should be an automatically applied command to invert each greyscale original image from negative to positive. This resulted in two uncommon situations in which negative digital images appear:
Complete Negative Microfilms
A very few of the microfilms were replacement films, and some of those were positive duplicates. If this was not caught by the technician, the universal invert command would change these images from the positives we preferred to the negatives we wished to avoid. This has happened, and we have corrected it wherever found. Here is a made-up example of a negative, then the corrected positive, of the same image.
Situations Where Negative Images Were Photographed
Also, there are a few situations where the originals actually are negative images. Most of these are photostats, negative photocopies made on a machine invented early in the 20th century. They found their way onto our microfilm because some of our rare books are incomplete. Through much of last century, upon request one special collections library with a complete copy of a title would make photostats of the pages missing from another library’s copy, so researchers at the second library would have access to a “complete” copy. These images were shelved next to the rare book in an envelope, and they were microfilmed with the book at the end of the microfilm until recent decades. They would appear on the negative original microfilm as positive images. In this case, it is entirely appropriate to find them as negative digital images, as in the above.
3. Digital Autocrop Problems
Here the autocrop function focused on the white page only in order to determine the top edge of the image, as the fold-out was so dark that it was more similar to the black background than to the white page. The resulting image is missing the top portion of the fold-out.
4. Automatic Digital Image Cropping Problem
This is actually quite understandable, if somewhat odd. When circumstances made it difficult to place the call number flag in or under the item being microfilmed, it would usually be placed next to the item. Here is an example of proper cropping of a flag placed beside the item.
However, it is easy for the autocrop function to decide that the item and the flag are separate images, rather than one image. This becomes more likely as the distance increases between item and flag in the original image, which is not a problem on the microfilm.
A second interesting situation appears when a black leather binding is microfilmed with a flag sticking out of the top. The resulting thumbnails look like this:
The actual images look like this:
5. Image skewing
This is a very strange problem, and not at all the same thing as item skewing. It must be due to something in the digitization process, but only because I can’t see how one could do this on a camera, while it would be easy in Photoshop. It can’t be film camera skewing because the vertical edges are still vertical, while the horizontal edges are skewed in some images but not in others. Also, the focus looks pretty good. Sometimes it disappears as the images progress, while other times it is found throughout the microfilm.
6. Image splintering
Another strange and very rare problem is what I call splintering. The Photoshop-corrected version is missing a strip of the page about one-letter’s width, but its symmetry proves that it corrects a single image that is splintered, not the result of a splice.
Conclusion: A Plea for Assistance
The wide variety of oddities we have found so far in the digitized microfilm from the Folger Microfilm Collection lead me to believe that many more examples lie patiently waiting to be found and puzzled over. Happily, most are innocuous. Enjoy those, with an insider’s feeling of superiority. If, however, you find one that causes real problems, such as 358 images of a book that are upside down, please let us know so we can correct them to make life easier for you and your friends and fellow explorers!
Note: Helpful definitions of microfilming terms can be found in “Components of Film Quality,” the last section of 3. Microform Terminology in Managing Microforms in the Digital Age, by Kitti Canepi, Becky Ryder, Michelle Sitko, and Catherine Weng, produced by the Association for Library Collections & Technical Services, a division of the American Library Association.
William Davis is Senior Photography Associate at the Folger. He has done many things related to imaging and cataloging there since 1993, including working closely with the microfilm collection, completely digitizing 19 First Folios prior to the 2016 tour, and advocating for shelfmark and pagination standardization, none of which have anything to do with his training in the history of religions in South Asia.