Vol. XVI, No. 3

Winter, 2004

The data from the investigation of the older propylon has recently been prepared for sending to be archived at the Archaeological Research Institute (ARI) at Arizona State University. Please note. The archival materials that were stored at the Archaeological Research Institute have been moved to the Archaeology Data Service, located in York, England. The ADS' website is at http://www.archaeologydataservice.ac.uk. Files concerning the older propylon project may now be found there and are no longer at the Archaeological Research Institute. The CAD model of the older propylon, along with related data files, had been available on the Web, but neither the model, the data files, the images, nor the notes made during the project had been properly put into shape for permanent archival preservation until they were prepared for the trip to Arizona. Like that infamous shoemaker whose children must go barefoot, Susan Jones and I had preached the need to archive excavation data but had assumed that our files were in satisfactory condition until we really had to contemplate sending them to another archive. Then we realized how much had to be done.

Before we began the process of preparing the files, we decided that it would be useful to note carefully how that process worked, how much time it took, what kinds of problems we encountered, and what expenses were involved.

It has turned out that the process was more time-consuming than expected but not necessarily more difficult.

One part of the process was not, strictly speaking, preparing the material for archiving. I made new recommendations for linking data tables to CAD models some time ago (see CSA Newsletter, Vol. XIV, No. 3; Winter, 2002, "Linking Text and Data to CAD Models" -- http://www.csanet.org/newsletter/winter02/nlw0201.html), but I had not fully updated the older propylon CAD model and data tables to reflect the new recommendations. That updating process was completed as part of this broader effort to prepare the files for archiving.

These were the required tasks:

• The CAD model had to be updated, as did the linked data files, to be sure that all met the most recent recommendations for models with linked data. The CAD files also had to be documented in terms of saved views, script files available, and so on. (The layer-naming system had already been documented.)
• All paper records -- field notes and drawings as well as correspondence and notes and drawings made after the completion of field work -- had to be scanned so that the archive could be completely digital, not part digital and part on paper. At the same time, each of the scanned documents had to be labeled, meaning that I had to try to decide how to label the content of each page, though there were some I had not looked at for decades.
• Many slides had been taken on site -- many of which were duplicates or near-duplicates. Those that had not been scanned previously for use on the Propylaea Web site had to be scanned, as did black-and-white photos taken for the photogrammetry work in 1989. Given the duplications, not all slides were needed; so the process included choosing the slides to be scanned.
• The data table documenting the slides -- information about date, film type, photographer, subject, etc. -- had to be expanded to serve the archival purposes. (Ms. Jones had already constructed the data table for internal use.) Here again, I had to provide subjects for each slide, a relatively easy task in the case of the slides.
• The documentation for the overall content of the collection had to be assembled and written. That meant specifying all files, all linkage between and among files, data types, file types, file histories, and so on. This was the most difficult part of the process, since it had to be full and accurate without dissolving into unintelligible computer-speak. (The result is available here -- see http://www.csanet.org/newsletter/winter04/OPrxivDoc.html so that readers may examine the documentation we deemed necessary. Readers' comments are welcome.) For instance, it is important that the history of a file be known -- not simply what software was used, but what versions and how the upgrading of file formats was carried out -- but that kind of information is hardly scintillating in text form.
• The copyright statement had to be updated, placed into the CAD model, and included in the documentation. It reads as follows:

  "Copyright, Center for the Study of Architecture, 2004.
  
  "CSA holds the copyright to this document/model. The right of any scholar to access and use the model for true scholarly purposes is expressly granted. In addition, CSA expressly grants to the Archaeological Research Institute (ARI) at Arizona State University the unlimited right to use the model for true scholarly purposes and to permit other scholars to access the model for scholarly use. Commercial use of any kind is only permitted with the written permission of CSA; commercial use includes any use for publication or public lectures that are not free to all."

  The documentation file also states that the copyright passes to ARI if and when CSA ceases to exist.

• We did not try to supply resource discovery metadata. That is, we made no effort to supply an isolated list of keywords, geographic terms, chronological terms, or other terms that might help a researcher locate these files, although we did describe the project carefully so that the terms could be taken from that description. We assumed that such data will be required, but personnel from ARI will look over the supplied information and then help us generate keywords for resource discovery or ask us to supply specific kinds of terms using their standards.
• The files had to be written to CDs (in this case 8 CDs) and sent off to Arizona.

The only costs other than personnel time and the minimal costs of the blank CDs were for scanning of slides and negatives. Since our earlier experiment had shown that better quality slide scans were achieved by sending the originals to Luna Imaging in California, that was the choice here. We sent 105 slides and 46 black-and-white negatives to them, and the cost of scanning was $630.00.

The bigger costs were for Ms. Jones' and my time working on the various pieces of the process. Ms. Jones spent a total of about 24 hours, including time dealing with the photographs (helping to cull them, expanding the data table that holds information about them, entering data, arranging for mailing, etc.), scanning and documenting the paper records, proofing various documents, and transferring files from one format/medium to another.

I spent more than 40 hours on this work. The lion's share of the time was spent on the documentation, a sixteen-page text file explaining -- we hope -- all the necessary information about the files. However, I also spent time on all the data tables, making sure that, for instance, all the entries were consistent and could be properly documented. I was also obliged to clean up the AutoCAD model in ways I had not anticipated. Since I had continued to use the file regularly, I had not realized how many little things needed to be changed for the sake of a naïve user. For instance, I had kept the model oriented so that the model's y-axis was not up, as it should be. Instead north was up, not a bad choice to be sure. However, the model was built with the y-axis oriented along the wall atop the steps leading into the upper courtyard of the older propylon, not to the north, and what AutoCAD calls the world coordinate system uses that orientation to determine what is normally up on the display. The model needed to be updated so that it opened with the orientation based on the wall; that is the orientation used to create the model; so it is the appropriate orientation for a new user. (I had even prepared a copyright statement that would show to the user on opening the file. Alas, it was oriented incorrectly as well and had to be corrected.)

Ms. Jones and I had a significant advantage in this process. We had talked at length about the kinds of things required and had written about the processes as well. For instance, we knew what had to be documented, and we knew that files had to be saved in multiple formats. In addition, we were familiar with the processes required to save files in various formats and with the formats needed. Finally, we did not have requirements placed upon us by the ARI repository. In short, we did not have to spend precious time getting/making a list of requirements. It is not possible to estimate the time that saved, but I would hazard a guess that we would have spent at least half again the amount of time on the project had we not already had a good idea of the requirements.

We also had the advantage that some of the documentation had been done when it should have been -- as the project was progressing. In particular, the layer-naming system had been thoroughly documented years ago, complete with the data table for the layers, with each layer name explained.

We had an even more important advantage. I did the original work. I built the model, took the photographs, constructed the data tables. Had this documentation process fallen to someone else, I shudder to think what questions would have been beyond answering.

Much of the documentation could -- and should -- have been done sooner. Not only would it have been more efficient to document the work along the way, the documentation would have been fuller and more accurate -- and the risk of finding it impossible ever to complete the documentation would have been greatly lowered. That is the key lesson to be learned from this experience. Documentation is critical to data re-use; waiting to undertake the documentation introduces unnecessary risks and costs.

Harrison Eiteljorg, II

To send comments or questions to the author, please see our email contacts page.

For other Newsletter articles concerning the issues surrounding digital archiving, consult the Subject index.

Next Article: Note-Taking with Total Station Survey Work

Table of Contents for the Winter, 2004 issue of the CSA Newsletter (Vol. XVI, no. 3)

Table of Contents for all CSA Newsletter issues on the Web