Using Edge Detection Algorithms to Search for the
Physical Remains of Roman Centuriation and Surveying

But the land surveyor is like a judge; the deserted fields become his forum,
crowded with eager spectators. You would fancy him a madman when you see
him walking along the most devious paths.
----Cassiodorus

It is well known that the remains of Roman Surveying throughout Northern Tunisia are the best preserved in the world, but the difficulties in geographically locating these areas has led to a serious lack of research and scholarship on these remains. The Romans, in the regions around Carthage, Dougga and Enfida, surveyed extensive areas, and left behind physical remains in the form of limites and field boundaries. I have recently begun studying these regions using both fieldwork and computer methods in an attempt to locate and map these important remnants of Roman colonization. In the figure below I have used an edge detection algorithm on a satellite photograph (Landsat) of the area around Carthage that allows for the enhancement of linear field boundary features and whose results can be statistically compared to the known forms of Roman cadastral surveys found in texts such as the Corpus Agrimensorum.





















[click on figures to enlarge]

Once the extent and size of these 'survey areas' has been determined a grid can be fitted and overlaid on a satellite photograph of the region in question. The figure below shows the area around the modern town of Enfida, Tunisia. The calculated grid lines up quite well with the current local path system around the fields and with the intersection known to have been the kardo and decumanus in Roman times.

Below one can see a GIS map that I produced showing several of the areas studied so far using these methods along with a grid over the approximate extent of the physical remains. More complicated algorithms using both Fourier and Radon transforms have also been used to locate and orient the regions shown. I have presented detailed results of this research at the International Conference on the History of Cartography held in Copenhagen in July of 2009, and will provide more on the mathematical details of this in a future post and publication.

Exploring Waldseemuller's World :

An International Symposium at the Library of Congress

For those of you who where unable to make it to the Library for our Waldseemuller Conference last month we have posted the complete conference on-line in four 2.5 hour parts... you can find the video webcasts at...

Part 1: Scholars and Scientists

http://www.loc.gov/today/cyberlc/feature_wdesc.php?rec=4568

Part 2: Exploring the Known and Unknown

http://www.loc.gov/today/cyberlc/feature_wdesc.php?rec=4567

Part 3: Sources and Texts

http://www.loc.gov/today/cyberlc/feature_wdesc.php?rec=4569

Part 4: Changes and Revolutions

http://www.loc.gov/today/cyberlc/feature_wdesc.php?rec=4566

Hope you all enjoy...

Hidden Toponyms:
Corrections to Place Names on Waldseemüller's 1516 Carta Marina

“A man of genius makes no mistakes. His errors are volitional and portals of discovery.”
--James Joyce

On sheet number 9 of the 1516 Carta Marina by Martin Waldseemüller there are two ornamental shields, one of which is covered by a cutout piece of paper of the same shape (figure 1). It has been thought that beneath the shield occurs a series of printed errata that pertain to corrections that may have been made to the map. The full transcription and analysis of these errata have until now however proved illusive[1]. The hyperspectral imaging accomplished last year at the Library of Congress[2] has provided the means with which to finally transcribe these errata and analyze their relationship to the 1516 Carta Marina map.

Transmitted hyperspectral images of the map sheets were taken at a variety of wavelengths in the near infrared and have allowed the viewing of the printed text through the pastedown paper overlay (figure 2).

The inscription shown in the figure consists of a series of changes that needed to be made to the map and has provided evidence that one of the sheets (number 6) is an earlier proof copy. The changes consist mostly of misspelled and misplaced city and region names.

Sheet 6 of the Carta Marina shows the coast of West Africa and does not contain the red lines drawn on the rest of the map by its owner, the Nuremburg mathematician and astronomer, Johannes Schöner (1474-1547). This particular sheet was also not bound into the Schöner Sammelband in the same way as the other sheets of the 1507 and 1516 maps and displays a different watermark from the others.

On errata list above imaged using hyperspectral techniques there are two errors, “seu brachi” and “Teopardi”, that remain uncorrected on sheet 6 while all the other corrections listed in the errata shield have been made to the map. The fact that sheet 6 remains uncorrected in relation to the errata list has led us to conclude that it was most probably a proof sheet. This conclusion of course contrasts with the previously accepted view that sheet 6 was printed later.

Schöner reproduced sheet 6 as a manuscript on vellum and had it bound into the Sammelband in the same way as the 1507 and other sheets of the 1516 map. His drawing shows the changes that are suggested in the errata list suggesting that he either had access to the list or to a later non-proof copy of the map.

This technique has also been used to look at the pastedown on which the text block of the lower right-hand corner of the 1507 map is printed on. What one sees coming through is actually the text from a palm reading manual from 1515 indicating that at least this part of the Library of Congress' 1507 map was printed after 1515.

[1] E. Harris, “The Waldseemüller World Map: A Typographic Appraisal”, Imago Mundi 37 (1985), 30-53. In Harris’s article she reproduces the first videocon image of the errata shield using transmitted IR taken by Dianne van der Reyden at the Smithsonian’s National Museum of American History. Van der Reyden is currently the Director of the Preservation Division at the Library of Congress. Although Harris and van der Reyden were able to determine that the text was composed of errata, no transcription was possible at the time.

[2] Imaging Team for the Waldseemüller maps at the Library of Congress consisted of Michael Toth, R.B. Toth Associates; Dr. Keith Knox, Boeing Corp; Dr. Roger Easton, Rochester Institute of Technology; Dr. William Christen-Barry, Equipoise Imaging; Doug Emery, Emery IT; Dr. Fenella France, Library of Congress; Heather Wanser, Library of Congress; Kenneth Boydston, Megavision Imaging.