Artwatch UK

Destroying archives

26 August 2013

Technological advances are often over-sold and deployed in haste. As Nicholson Baker famously showed in his book Double Fold ~ Libraries and the Assault on Paper, countless books, magazines and newspapers were destroyed when microfilm seemed (falsely) to be a better, more durable, more economical means of storing their “information”. The BBC discarded much irreplaceable historic material which, having been shot in black and white, was held technically obsolete on the arrival of colour productions. As we reported on February 28, 2012 (“Shedding archival records at the Tate and the Victoria and Albert Museum”), the Paul Mellon Centre for Studies in British Art had recently received a phone call from a Tate employee who said “you might like the curatorial photo archive because we’re about to throw it on to a skip”. We subsequently learnt that the threat to archival material was more widespread and that it was being strongly resisted. Technical advances and their attendant risks are not abating. Here, the painter and ArtWatch UK Journal’s picture/photography analyst, Gareth Hawker (- see his post of 10 January 2011 on photography in museums), discusses some of the dangers posed to archives by breath-taking but commercially driven and insufficiently examined technical developments in digital photography.

Gareth Hawker writes:

The benefits of digitising archives can seem immense: the archives become easier to index and retrieve than the original documents, and copies may be sent anywhere in the world almost instantly. These advantages can appear so dazzling that the risks of digitising may be disregarded, especially by institutions which lack the funding and the expertise which the major museums can call upon. For those with low budgets and little experience in digitising their archives, several groups have issued guidelines – the “Western States Digital Imaging Best Practices” [Endnote 1] being among the best known. These practices have been taken as the basis for a number of instruction manuals, notably one written by Jim Kennedy [2]. He provides plenty of useful advice but, while he does mention the risks involved, the bulk of his manual describes ways in which an image may be manipulated. An inexperienced archivist may get immersed in this part of the book and, in his enthusiasm for manipulation, throw away the original file.

It is true that the original archive file may not be suitable for all uses, for example a publisher may wish to enhance the image – perhaps by increasing contrast and removing blemishes – so that it would look better in a book or on the Internet. The resulting picture can look quite different from the archive image, but a researcher should always be able to track back to the unaltered original and check through any changes that may have been made. Ensuring that this is possible is known as maintaining image integrity. Digital images are intrinsically more proof against tampering than analogue images, but only if they are stored with the audit trail which records the changes, if any, which have been made to the original digital file. The construction of an audit trail is described in the Adobe document, “Digital Image Integrity” [3]. However, among the procedures which the “Western States Digital Imaging Best Practices” document lists as ‘best practice’ it includes deleting the original file, and keeping only the manipulated version as a ‘master-file’. This invalidates the file as an archive, especially if the original photographic print or transparency has also been thrown away.

Jim Kennedy writes:

“Best practice is:

(a) to make a master image that has tone and color carefully adjusted to correct fading and exposure, or (b) to make a master image that represents the tone and color for the physical condition of the item at the time of digitization without correction of fading or exposure. The choice depends upon the goals and resources for the project, with the second option requiring more extensive resources to create and maintain large files that may never be used and include reference targets when possible.

Both types of master images could be included in an archive. …”

Clearly choosing option (b) is vital for a serious image archive. Relative to the cost of time spent in scanning and filing, the cost of storage on a disc or drive is tiny; but throwing out the original file can cause confusion for ever after. The archivist who is pressed for time need not make any of the manipulations which the guidelines suggest. These could be postponed until someone wanted to adjust a copy of the archive file for a specific purpose, leaving the original untouched.

The task of preserving the new digital records presents new problems. Hard drives and discs become corrupt with the passage of time, so the data on them needs to be transferred to a new set of drives or discs before it is lost. The entire archive needs to be transferred frequently – every two to four years according to some authorities. There is a danger that someone may forget to transfer the data and that it will all be lost. This is a good argument for keeping the original, non-digital documents, but one which is sometimes overlooked. For anyone thinking of digitising a collection, David Saunders’ chapter on the subject of preserving records, “Image Documentation for Paintings Conservation” in Conservation of Easel Paintings (Eds. Stoner and Rushfield), provides a summary of this and other considerations which would be worth bearing in mind.

One consideration is the quality of the digital files themselves. The resolution of a digital photograph or scan is likely to be far lower than that of an old-fashioned photographic print – even one of poor quality. The amount of data lost if the original print or transparency is thrown away is incalculable. In addition, new techniques may be able to retrieve even more data from the originals than was ever imagined. Destroying the original documents will close off this possibility forever. Keeping the print-out of a digital document (‘hard copy’) may be advisable, but is an inadequate substitute for keeping the original, pre-digital document.

As an indication of the types of falsification that the writers of the ‘Western States Digital Imaging Best Practices’ consider acceptable, see the examples at right. A researcher looking at a ‘master image’ would have no way of distinguishing between what was a true record, and what had been doctored.

These doctoring procedures would invalidate photography as a means by which to examine how paintings had changed over time. This would represent a disaster for historians, and a blessed relief to any restorer who wanted his blunders to be forgotten. Restorers themselves rarely seem to look critically at ‘before’ and ‘after ‘photographs of the paintings that they work on, while museums often keep only haphazard photographic records of the works in their collections. The Rembrandt [4] and Raphael [5] databases give some idea of how incomplete these records may be. Perhaps this directionless attitude to record-keeping derives partly from the restorers themselves, who do not often attach much importance to comparing ‘before’ and ‘after’ photographs. Most restorers prefer to monitor their own work according to what they see through the microscope, informed by their own experience and training – but without any objective standard against which to measure the result of their actions. Few restorers outside the major museums take high-resolution ‘before’ and ‘after’ photographs of the paintings they work on, though they may take snapshots which are low in resolution, unevenly lit, and inaccurate in colour. Not having accurate ‘before’ and ‘after’ photographs available – and, when they are available, not being practised at examining them – most restorers have little opportunity to assess the extent to which their work has damaged a painting.

Thus the importance of photographs may be underestimated, and the contribution of the archivist undervalued. It is essential that the best possible photographic records be made and maintained if any objective assessment of changes to a painting’s appearance is to be undertaken. An archivist may do well to consider keeping original, pre-digital documents, and resisting the temptation to become completely dependant on the computer.

Gareth Hawker







Comments may be left at:

Printer-friendly PDF version of this article

Above, Fig. 1: Unknown Lady (detail) by H. T. Wells, R A c 1880.
Left : Archive file Right: Master file
The ‘Archive file’ records the data in a verifiable way, while the ‘Master File’ shows the data adjusted in a non-verifiable way according to the skill and judgement of the editor.
Above, Fig. 2: This archive file includes a target. This is to help the photographer to capture colour and tone accurately, and to enable the researcher to check that he has done so. (In this example an inspection of the chart reveals that the lighting was slightly uneven and the camera not quite perpendicular to the canvas).
Using a spectrophotometer, the manufacturer will have taken readings from each patch on the actual chart (in this case a piece of photographic paper), and recorded them. To check the readings on the archive file, the researcher can open it up in an image manipulation programme and click on each patch in turn with the eyedropper tool. The more accurate the photograph, the closer will the eyedropper readings correspond with the manufacturer’s. It is a simple task for the researcher to check colour-accuracy using a Wolf Faust IT8 target (as shown here) because the readings are provided with the target. They may also be retrieved from the Internet by quoting the ‘Charge’ code at the bottom right of the target. However, some other commonly used targets, such as the Kodak Colour Control Patches and the Macbeth/Xrite 24 Patch Colour Chart, are neither standardised, not are they (usually) provided with spectrophotometer readings. If the original photographer has not taken and recorded the correct readings of these targets, using his own spectrophotometer, the researcher will have no way in which to verify colour-accuracy.
Above, Fig. 3: The contrast increased slightly and the colour balanced according to an educated guess on the part of the editor. The process used to achieve a pleasing image is sometimes called, ‘colour-correction’ even though it is concerned with making the colour inaccurate. (‘Colour accurate’ means correct according to objective standards, while ‘colour-correct’ means pleasing to the operator who made the adjustments – not conforming to any objective standard). The series of images shown here illustrates the beginning of an attempt to show what the painting might have looked like before its varnish deteriorated and became dirty. The grey scale on the target shows what this adjustment has done to the neutral greys.
Above, Fig. 4: The contrast has been increased even more. The greyscale shows how the tones have been compressed (made almost the same) in the lights and the darks, while the differences between the tones in the middle of the scale has been expanded (increased). This makes the picture look more acceptable in the eyes of many viewers. Again the greyscale on the target indicates the nature of the adjustment that has been made. With reference to this target a researcher could conceivably reverse these procedures, at least within the mid tones, but the next step would be impossible to reverse.
Above, Fig. 5: Using a complicated technique, some of the yellow patches of varnish have been made less obtrusive. This process would be impossible for a researcher to reverse, unless there was a full record of the editor’s actions in an audit trail.
In addition, using the ‘clone stamp tool’, some of the spots of dirt have been covered over by copying tiny parts of the image and pasting them on top of the spots.
This sort of retouching, done thoroughly, could take many hours, but perhaps this demonstration is sufficient to indicate the possibilities. In many case it is impossible to see where adjustments have been made: they are untraceable without an audit trail. This shows why it is so important to preserve the archive file, even when the manipulated file may look more pleasing.
Above, Fig. 6: A review of the manipulations described above.
1 Archive File 2 The contrast increased slightly and the colour balanced 3 The contrast increased even more 4 Some of the yellow patches of varnish made less obtrusive 5 Some of the spots of dirt covered over
Above, Fig. 7: Archive file – Face only
Above, Fig. 8: Master file – Face only
Many viewers would find the Master file more pleasing to look at. It may possibly give a better impression of what the painting looked like when it was new. However the Master file is the result of many arbitrary decisions on the part of the editor: unlike the Archive file, it cannot be regarded as a true record of the painting.
Click on the images above for larger versions. NOTE: zooming requires the Adobe Flash Plug-in.

Comments are closed.