|Making Duplicate Films for Scanning:
Alta California and San Francisco Call
Library Photographic Service (Preservation Department) at University of California, Berkeley has the limited role in the California Newspaper Project of reproducing master negatives for purposes of scanning. The microfilm section of this service is accustomed to, and well-equipped for, duplicating difficult old camera masters, as the institution holds large collections of film materials dating back 60 years and more, and requests for copies of these films are received with considerable frequency from institutions and individuals around the world. The product created for CNP is 35mm duplicate negative film, on a modern, lower-contrast, direct-duplicate intermediate film. This film stock has an orthochromatic, gelatin-silver emulsion on a polyester base. A "vintage" contact printer especially well-suited to the work is employed, and the exposed film is subsequently developed in a continuous-strand, deep-tank processing machine. Metadata pertaining to certain technical aspects of both film generations (master and duplicate) must be collected as part of the process. These data include such information as reduction ratio, date of production, resolution, transmission density readings (20 altogether), minimum densities, density averages, etc. The finished product is put into approved enclosures and shipped to the scanning vendor.
The camera negatives:
The original film negatives were stored spliced together on 1000-foot rolls (on plastic film "cores"), which were in turn housed in metal film cans. Production dates are not stated on the film, but are estimated to have been between 1960 and 1970, the estimate being based on their characteristics:
The majority of the material was filmed one page per frame, in the "1A" format, a good decision on the part of the filming agent if it was not mandated by the ordering institution. Some of the newspapers were filmed in the "2B" position, however, and a significant number were done in what we are calling "one-half B", which is to say that half of each sheet is filmed in each frame, in the B position - sometimes referred to as "comic" or "landscape", the length of the documents running across the width of the film. In the aforementioned A-position method, the length of each document - or pair of documents - runs down along the length of the film.
It is apparent that the cameras used in the original filming were Recordak Model C machines. There is abundant evidence to confirm this supposition. This model's swinging arm device, which incorporated the focusing aid and light meter, was inadvertently left in the field and photographed in some instances, and there is clear evidence of the use of the shifting or oscillating copyboard which came with these machines.
The Model C design was an electro-mechanical marvel, extremely durable, facilitating high reductions for imaging large originals. These cameras were fitted with an earlier generation of the Ektar copying lens, however, which is inferior to the later Ektars and to the later, comparable designs from other manufacturers. Model C's also lacked the improved film pressure plate of subsequent models, although these platens may be replaced. The focusing method was rather crude and imprecise in practice, and lacking in visual or instrument focus confirmation. The consequence of these shortcomings is perhaps the most important characteristic for purposes of the CNP project: medium to low resolution. The determination of these resolution levels is necessarily subjective, since test objects (resolution charts) were not filmed by the vendor. Inspection of reproduced materials under magnification reveals soft and diffuse edges, spread or irradiation, and a tendency to closure of small spaces within characters. As expected, resolution loss tends to be in more-or-less direct relation to reduction; that is, the greater the reduction the greater the loss in definition. The reduction ratios of the project films duplicated to date run from 13.5: 1 to 21: 1, with the lesser reductions having produced the medium levels of quality and the greatest ones the marginal or low levels. Generally speaking, in a good film reader the quality does not descend to the level of illegibility, but rather to the level of what may be described as "legible with difficulty" or at worst "decipherable". I estimate line-pair resolution to be on the order of 90 line pairs per millimeter which, again, and depending on character size, is adequate at the lower reductions and marginal or problematic at the higher ones. For purposes of comparison, I can state that the current cameras in Library Photo Service (preservation microfilm lab) are adjusted to reach around 150 line pairs per millimeter, and some manufacturers and filming agents claim even higher figures in their advertising literature.
The scanning agent for the duplicate films (OCLC) indicates that the duplicate films are scannable, and as of this date we await the results in terms of OCR error, on-line presentation, etc. The splices in the film also help to date the filming. These are rather thick overlap splices, done with liquid adhesive, almost certainly made with the Griswold brand splicer that was used most every where at the time. Generally these splices remain strong and are sufficiently distant from the images to prevent imperfect contact in film duplication. Where tape splices were used significant problems have resulted, and these will be discussed below.
The old camera negatives for Alta California and The Call suffer from the following additional problems with respect to their film and digital copying:
The first of these defects results in film density ranges that make reproduction difficult, as well as, in some cases, individual images that were effectively ruined by gross underexposure or overexposure. The incidence of this latter problem is relatively low, and credit must be given to the microphotographers for getting acceptable exposures over the course of a large project that included many thousands of pages of difficult source documents filmed on a production basis and on older equipment. Mismatching refers mainly to images added to the reels by splicing in refilmed sections to correct errors or to add pages found later. Process control was more challenging than it is now, with the result that the spliced-in sections do not match surrounding material in terms of image characteristics. In these cases the old-style contact printer employed in the photo service is useful, as it may be programmed for many different contact-printing exposures within a reel. This programming is done with the goal of making scanning more straightforward, although we have little doubt that some adjustments to thresholding are still needed on this difficult material. The scanning vendor asks that negative "background densities" fall between log .9 and 1.2, and every effort is made to achieve this, although the first-generation densities range from about .4 to nearly log 2.0
At the outset of the project OCLC was presented with three alternative versions of a single roll to enable them to choose the one with the best characteristics for their process. Lower contrast was preferred, with a "base-plus-fog" density of around .10. The original film suffers from two types of deterioration: the degrading of cellulose acetate support - "vinegar syndrome" - and reductive-oxidative damage, familiar to many librarians as "microdots". The use of test papers and visual inspection indicate that the former of these two problems is not advanced in the films, although the tell-tale acetic acid odor is detectable in some. The second defect - redox, for short - is not widespread but is severe in some places. By way of review, redox occurs when migrant silver ion is created by oxidant attack typically catalyzed by heat and moisture, so that a colored, non-image form of the silver is deposited in a new place. In the finely-divided silver of close-grained copying films, this re-deposit often takes the form of tiny dots - "microdots". Most of the dots in the films under discussion are confined to image edges outside the text area, however there are a few frames that are quite decimated by the phenomenon. Where tape splices are present, direct contact with the adhesives has in some cases triggered broad areas of damage, apparently "wiping out" up to two-thirds of the frame in contact with the tape. Here the above-mentioned orthochromatic emulsion has been useful, since the color sensitivity of this material makes it sensitive to some of the wavelengths passed by the blighted, colored areas, and an overall improvement is noted in the new copy. The effect of enhancement has been quite dramatic in some cases, however there is no question that some dots will record in the scans.
It should also be mentioned that some of these masters were used as reading copies and are scratched and abraded as a consequence.
The final limiting factor noted above is the physical condition of the newspapers at the time of filming. We can assume that many of the sheets were quite brittle, and in some cases there is significant text loss owing to pieces having broken off. Some sheets have tape repairs covering text. Patches of faint imprint occur, and it is difficult for photographic copying to capture adequately the entire range of values on pages affected in this way.
It should be noted here that some years of Alta California were printed on such large sheets that the filming agent found it necessary to film these pages in halves, which, while it introduces no special film duplicating problems, raises issues for OCR and web presentation.
From the limited perspective of this paper it is impossible to offer meaningful conclusions about the project as a whole. At the time of this writing we have seen only a few test scans of the material in this facility. However, the following points from the perspective of a preservation microfilm lab may be germane to the general discussion:
Additional remarks in re "the left-hand side" problem
In the report of 6/06 we neglected to mention what has already been identified as a problem for scanning on Alta California and The San Francisco Call. While the scanning agent will undoubtedly treat this problem in discussion, explaining how it was addressed, it can be described here from the standpoint of a master negative defect.
We refer to the phenomenon of higher densities and low definition in the frame area corresponding to the left-hand side of the newspaper sheets, relative to filming in the 1A position.
The density difference between the right and left-hand sides is often found to be on the order of one stop or more (log .3 or greater), and the separation between text and background is noticeably poorer in the left hand text column (s). The terms "soft", "fuzzy", or "washed out" apply.
After eliminating the duplication process as a significant contributor to the problem we found it difficult to pin down the cause of this defect. It may be a compound problem: uneven lighting across the copyboard is always an issue with large sheets, however the appearance of the material is more suggestive of poor film-plane flatness, lens aberrations, internal reflections in the optics or the camera, etc.
Regardless of the cause, there is little that can be done photographically by way of correction. Printing-lamp adjustments could be made to give more exposure to one side than the other, but this is a trial and error process, involving a lot of testing and densitometry, and it would leave our contact printer out of adjustment for the majority of the duping done here. Ultimately this step would likely not solve the problem in any case, as the higher local density is only a part of the problem, and "hot spots" exist in many parts of the frames, due to weak imprint, etc., but without the accompanying soft focus. Digital enhancement/ sharpening seems indicated, but these are decisions made after the material leaves our facility.
The Regents of the University
of California. All Rights Reserved.