goto: The NEUMES Project homepage Neumed and Ekphonetic Universal Manuscript Encoding Standard  

Digital photography of manuscripts. Image source: DIAMM. Digital Photography

The Chapter of Worcester Cathedral has given kind permission [1] for the NEUMES Project to use images of the Cathedral's manuscript folio F173, pages 6r and 6v. These are among the first high-resolution, high color-depth, digital images of Western unheighted {def.} notation in existence. The quality of these images is exemplary for archival and scholarly use. We hope that other institutions will photograph their chant manuscript holdings at such high quality.
Worcester Cathedral, southwest view
  Worcester Cathedral traces its roots to a Benedictine Catholic monastery that was founded in Worcester, England, more than a thousand years ago.
Above: southwest view

These images were photographed by DIAMM in 600dpi {def.}, 24-bit color resolution (approximately 16 million distinct colors).[3] The original images are archived in uncompressed TIFF {def.}, and are rather large—approximately 109 megabytes each. (Archival images can also be stored without loss of detail in the PNG {def.} format, at a savings of approximately 45% in file size.) Our "Worcester Cathedral Fragments" presentation (below) uses JPEG image format, which sacrifices image detail but allows smaller files for transmission via the Web. A few of our image excerpts are minimally compressed, however. Our 'sample detail' images are at full scale and full color-resolution from the original image files.

  Worcester Cathedral fragments See, our presentation of the Worcester Cathedral Fragments


1. Purpose

Our purpose in presenting the Worcester images is to show the result of high color-resolution (24-bits per pixel) digital photography, which we believe ought to be adopted by manuscript-holding institutions for electronic archiving of neumed manuscripts. As permanent records, and as resources for scholarly study, images produced by this method are vastly superior to the common practice of microfilm for archiving of manuscripts. For example, compare our Worcester Cathedral 'sample detail' photographs (above) to this microfilm image: both film photography and digital photography may use high-resolution photography (viz., a large number of photographic pixels per square inch of source document), but microfilm images are crucially different in their depth-of-color.

Based on research results by others in 'digital restoration' of manuscript images, and our preliminary tests in digital character recognition, we believe that the 24-bit color depth is a crucial factor. Pixel resolution is, we believe, somewhat less critical: source imaging at 600dpi or 400dpi seems acceptable in most cases (although digital photography is capable of going to much higher resolutions). Sharp image focus also is crucially important. A standard color-reference bar for color correction, and metric rulers must be included in all images (see, the full-page Worcester Cathedral images, for instance). The financial cost of photographing manuscripts is high, and libraries justifiably wish to keep the handling of these artifacts to a minimum; for these reasons it is very important that, if an institution undertakes to photograph a manuscript, the photography should be of sufficient quality that the needs of researchers will be satisfied over the long term.

Of course, the advent of high color-resolution digital photography is a recent technological advance, and no one would dispute that earlier, film-based photography produced valuable resources for research scholars. Among these earlier efforts, the most important has been the on-going work of the Benedictine monks at St. Pierre de Solesmes Abbey, France, in publishing source images in multiple volumes of the Paleographie Musicale [4]. Digital images in high color-depth (as opposed to digitally-scanned images from black-and-white microfilm or greyscale photos like those in the Paleographie Musicale) are needed for various reasons. The factors of most concern to us are as follows.
  1. They enable discovery of hidden details that are obscured by ink bleed-through, dirt, overlaid text, and so on (see, examples and discussion[3]).
  2. The capacity to computationally color-separate the ink used for neume notation from the ink used for chant text (as well as from 'background image-noise') will, we believe, be crucial to success in optical character-recognition of early sources.
  3. They enable study of paleographic features without in-situ inspection of source artifacts.

2. Technical Requirements

The following 'Check-list for ordering images' was supplied to the NEUMES Project by Dr Julia Craig-McFeely of DIAMM.[3]
  • Specify that you want High Resolution (and say specifically 400 dpi at real size or higher).
  • Uncompressed TIFF (or PNG) format both at capture and in delivery; NO JPEGs in the workflow.
  • 8-bit (or 24-bit depending on library protocols for describing bit-depth) RGB colour minimum (this is the standard setting); 16-bit (or 48-bit) creates much larger images but also stores more colour information (48-bit images can be opened in Photoshop only using the CS version).
  • Colour profile of capture device embedded in the image.
  • Industry-standard colour, grayscale, and size scales photographed beside the image but not touching it.
  • White and black points adjusted to avoid clipping at the point of capture (not afterwards).
  • Image sharp (i.e. in focus) to the finest level of detail.
  • NO unsharp mask applied during capture.
  • NO unsharp mask applied after capture.
  • NO rotation, deskewing, reshaping, levels, colour or exposure adjustment after capture (i.e. no adjustment of any sort after capture).

3. Extent of Sources

Today no one knows how many manuscripts containing neume notation have survived from the Middle Ages. John Caldwell (Faculty of Music, University of Oxford) estimates that, "If everything is counted, up to say 1500 [C.E.], I am fairly certain that [the number of books containing chant notation] would be approaching a six-figure number" [5] (as an educated guess, on the order-of-magnitude of 100,000 books). Similarly, University of Toronto musicologist Andrew Hughes estimates that the number of liturgical books {def.} surviving from the Middle Ages reaches well into six figures, and, of these, many contain chant notation [6].

Each book (or, codex) can be a compilation of materials (possibly from diverse historical periods and scribal origins), and some codices might contain only a small amount of neume notation. If we consider a 'document' to be an individual chant, then the total number of documents containing neumed chant notation is likely more than one million. Significantly, Hughes states that, "A minute number of these are catalogued at all adequately, and many fewer [are] known in detail" [7].

As a case study, we are interested in cataloging the Western chant manuscript holdings of the University of Oxford.

  See our working notes on neumed manuscript holdings in Oxford University. This may be of general interest as it shows the type of problems one encounters in enumerating sources.  

A few of these manuscripts that are of first-order importance to scholars we hope will be undertaken for 24-bit color digital photography by the Oxford University Libraries Imaging Service (OULIS) for inclusion in the online Oxford Digital Library (ODL) as public references. [See Early Manuscripts at Oxford University, which currently has some high-quality images of later (non-neumated) music manuscripts online.] Furthermore, we would hope to transcribe such photographed sources to NEUMES/NeumesXML data as a pilot initiative to demonstrate the integration of images and transcriptions as a unified, online reference.

The vast majority of extant neumed documents were written in square-neume notation with a four-line 'staff', and many of them are essentially duplicates. Early sources (e.g., those containing unheighted notation) are typically unique specimens: they are much more rare and individually important to scholarly study.

[1] Letter of permission on file.
[2] Source of photograph undocumented; presumed public-domain. Modified by L. Barton.
[3] See the DIAMM website for information on the photographic methods used.
[4] The Monks of Solesmes, Paléographie musicale, (Sablé-sur-Sarthe, France: Abbaye Saint-Pierre de Solesmes, 1889 – ).
[5] John A. Caldwell, private communication.
[6] Andrew Hughes, private communication.
[7] Andrew Hughes, Late Medieval Liturgical Offices, (Toronto: Pontifical Institute of Medieval Studies, 1996), p. 2.
•  Masthead image of field photography, courtesy of DIAMM.

Protected by law under one or more of the following copyrights:
Copyright 2005-2007, The University of Oxford.
Copyright 2003-2007, Louis W. G. Barton.
Copyright 2002-2003, The President and Fellows of Harvard College; contains software or other intellectual property licensed from Louis W. G. Barton, copyright 1995-2001 by Louis W. G. Barton.