23 newly discovered fragments of the Forma Urbis Romae
(and our first match)

Photographs and text by Marc Levoy
Additional photographs by Linda Cicero
June 13, 2001

In June of 1999, while we were scanning the 1,163 fragments of the Forma Urbis Romae (also called the Severan marble plan), archaeologists unearthed a previously unknown fragment at an excavation in the Roman Forum. Although they weren't looking for a piece of the plan, they also weren't surprised to find one, given the map's general dispersal after the fall of Rome. Although the fragment was small (12 cm long), it had been fifty years since one was discovered, so the find was important. Knowing that we were digitizing the other fragments, they rushed the new one to our temporary laboratory in the Museum of Roman Civilization, and we scanned it. A photograph of this fragment is shown at left.

Much to our chagrin, soon after we packed up our equipment and left Italy, 30 additional fragments were unearthed at the same site. At left is an aerial photograph of the historic center of Rome. At right is the Colosseum, at bottom center is the Circus Maximus, and at left is a curve of the Tiber River. The next image shows a closeup of the area enclosed by the yellow rectangle.

This photograph shows a portion of the Imperial Forum south and east of the Curia (the large block to the left of center). A yellow circle marks the location of the room in which the Severan marble plan was hung. Although the room is gone, the wall on which the map was hung still exists; it is now the exterior wall of the church of SS. Cosma e Damiano. The place where the new fragments were found is marked with a green circle. (Note that although the fragments were found in 1999, the photograph dates from 1990, so the location is only approximate.)

After restoration and cementing together of obviously matching pieces, the archaeologists were left with 23 new fragments of the Forma Urbis Romae, of which only one was contained in our database. Now what? Should we bring our 4 tons of computer and scanning equipment back to Italy? Impractical. Should we leave our database incomplete? Unthinkable. Fortunately, the Italians agreed with us on both points. So, with surprisingly little fanfare, the Ministry of Culture agreed to send the newly discovered fragments to Stanford for scanning. Antiquities rarely leave Italy, and newly discovered (and therefore unstudied) artifacts are particularly precious, so planning the expedition took time - two years to be precise. Here's what happened next.

On May 22, 2001, a specially built crate containing the 23 new fragments arrived by Boeing 747 at San Francisco Airport. After being escorted through customs, it was convoyed down to Stanford by a company that specializes in moving cultural artifacts. Where to store such a precious cargo? My office, of course! (It was rekeyed for the occasion.) Watching over the crate from the top of my display case is the left eye of Michelangelo's David, a momento of the Digital Michelangelo Project.

Note the seal on the side of the crate in the previous image. It's the logo of Lloyd's of London - the same people who insure steamships, satellites, and Springsteen's voice. Note also the two padlocks on the lid of the crate. The crate was accompanied from Italy by Laura Ferrea - an archaeologist with the Capitoline Museums in Rome, who carried the only key.

The next morning, we opened the crate. It's a marvel of Italian design, consisting of several hand-crafted wooden cases, padded within and without, and fitted to slide into the crate with nary a millimeter to spare. Each fragment was individually wrapped in tissue paper and nestled in its own custom-cut bed of foam. In this image, PhD student James Davis (at left) unwraps the first fragment, while Marc Levoy and Laura Ferrea share a light moment. (Photo by Linda Cicero of the Stanford News Service.)

Things get serious when we start handling the fragments: two hands at all times, eyes focused on the artifact. By wearing white gloves, we minimized the transfer of oils from our hands to the marble surface. (Photo by Linda Cicero of the Stanford News Service.)

Fragments larger than 20 cm in length (about half of them) were scanned on the Stanford Large Statue Scanner, the same one we used to scan Michelangelo's statue of David. At right is a closeup of a scan in progress. The laser stripe sweeps across the fragment surface at 1 cm per second. By analyzing the changing shape of the stripe as seen from the side by a precision video camera, the hardware can estimate the three-dimensional shape of the surface, at least along the path of the stripe. The resolution of the 3D data is 1/4 mm.

Smaller fragments were scanned on the Cyberware Model 15 scanner. To capture all sides of the fragment, while at the same time keeping the laser roughly perpendicular to the fragment surface (to maximize quality), we had to turn the fragment about 20 times, sometimes propping it at odd angles. The resulting 20 range images will be aligned and merged offline to produce a single 3D model for the fragment.

As we did in Rome two years ago, we separately photographed each fragment using a color digital still camera. Although the two floodlights facing the camera seem bright, the small but focused spotlight standing between the camera and the fragment is actually brighter. It is also placed so as to cast a raking light across the fragment surface, a lighting configuration frequently employed when photographing incised stone artifacts.

Here are photographs of some of the fragments, shown at the same scale. (The largest fragment is 40 cm long.) The leftmost fragment (fn8) seems straightforward: rows of shops lining a network of streets. The "V" symbol near the bottom of the photograph is one of the plan's standard symbols for a staircase. The middle fragment (fn44) shows similar rows of rooms, but the spaces they open onto are walled off from one another, so they are probably not streets. There was also much discussion about the rectangle of columns (or trees) at the top. The rightmost fragment (fn9) is one of the most remarkable on the plan. It shows shops lining a street, but they are separated from the street by what appears to be covered arcades. Note also that the street is painted red. This is the only known example of a colored street on the plan, and it poses new questions about the purpose of the plan. (See the addendum below for a proposed placement of fragment fn9.)

After three days of hard work, the fragments have been scanned and photographed. They now lie safely repacked in their crate, awaiting transport back to Italy. As if this week weren't busy enough, we had also just finished carving and breaking the Forma Aedificii Gatesensis. So on Friday evening, the team retired to Evvia in downtown Palo Alto for dinner. Clockwise from lower-left is Marc Levoy, Laura Ferrea, Natasha Gelfand, Steve Marschner, Jennifer Trimble, Nicolas Scapel, Sean Anderson, David Koller, Szymon Rusinkiewicz, and Tina Najbjerg.

What's next? We'll align and merge the many range images we acquired for each fragment using the software pipeline described in our SIGGRAPH 2000 paper. We'll separately correct and mosaic together the color photographs. Although it wasn't our original intent, we are considering mapping the color photographs onto our 3D models. Although they accurately cover only the top and bottom surfaces, these photographs nevertheless provide important information about marble veining, weathering, and as we've seen, coloration of the fragments. We'll then combinine these 23 new models and photographs with the other 1,163 to produce an expanded database of 1,186 fragments. Finally, we'll try to solve the jigsaw puzzle (see below).

Addendum - our first match

written March 23, 2004

fn9 and 351 Matched by the computer Re-opening the crates They fit! Circus Maximus

It's been two years since we scanned these fragments. In that time, we have built a publically accessible web-browsable database containing 435 of the 1,186 map fragments. Unfortunately, this release of the database does not include the 23 new fragments described on this web page, pending their publication by archaeologists of the the Comune di Roma.

Meanwhile, Stanford PhD student David Koller has been developing computer algorithms to find matches between the map fragments and to identify their locations in the ancient city. We've tried several different algorithms, with varying degrees of success. In one of these, we manually annotate each incised line as it leaves the broken edge of a fragment, marking its position, orientation, and type. In the leftmost pair of images above, red denotes a wall separating two stores, green denotes the storefront wall, and so on. Then, the computer compares the spacing, angle, and type of these annotations, reporting the matches it finds. In February 2003, David found his first match using this algorithm - between the newly discovered fragment fn9 (described earlier) and fragment 351 - a small, unidentified fragment from the original set of 1,163. This match was so close that the fragments fit tightly together, as shown above. This "marble lock" was dramatically confirmed during a trip to Rome in June 2003, where we reopened the crates and examined these two fragments. In the second-to-rightmost picture above, the two fragments sit re-uinited in the foreground. Seated on the table at left is archaeologist Riccardo Santangeli Valenzani, who first discovered fragment fn9 during an excavation in the Roman Forum. Standing behind him (and smiling broadly) is David Koller.

David has also identified this pair of fragments as belonging to the Circus Maximus (think chariot racing), and tentatively located them along the stadium's eastern side. His proposed placement is shown in the rightmost image above. This image is a photographic composite, created from our high-resolution color photographs of the individual map fragments. If this placement is correct (and scholars believe it is), then the colored street on fn 9 is the famed Imperial Way, where the Roman army marched in triumphal procession following military victories. The horizontal white line in this image depicts the boundary between two of the roughly 3 x 6 foot marble slabs into which the map is divided. The vertical gray lines on the image shows how the Imperial Way continues upwards from fn9 and 351, across this slab boundary, and past other, previously identified fragments, of the Circus Maximus.

In the year since his first find, David has found 20-40 more matches (depending on how many you believe). Most were found using the annotation matching algorithm described above, but others were found using different approaches. These matches are described in this paper, to appear in Bullettino Della Commissione Archeologica Comunale di Roma in 2005.

levoy@cs.stanford.edu

	In June of 1999, while we were scanning the 1,163 fragments of the Forma Urbis Romae (also called the Severan marble plan), archaeologists unearthed a previously unknown fragment at an excavation in the Roman Forum. Although they weren't looking for a piece of the plan, they also weren't surprised to find one, given the map's general dispersal after the fall of Rome. Although the fragment was small (12 cm long), it had been fifty years since one was discovered, so the find was important. Knowing that we were digitizing the other fragments, they rushed the new one to our temporary laboratory in the Museum of Roman Civilization, and we scanned it. A photograph of this fragment is shown at left.
	Much to our chagrin, soon after we packed up our equipment and left Italy, 30 additional fragments were unearthed at the same site. At left is an aerial photograph of the historic center of Rome. At right is the Colosseum, at bottom center is the Circus Maximus, and at left is a curve of the Tiber River. The next image shows a closeup of the area enclosed by the yellow rectangle.
	This photograph shows a portion of the Imperial Forum south and east of the Curia (the large block to the left of center). A yellow circle marks the location of the room in which the Severan marble plan was hung. Although the room is gone, the wall on which the map was hung still exists; it is now the exterior wall of the church of SS. Cosma e Damiano. The place where the new fragments were found is marked with a green circle. (Note that although the fragments were found in 1999, the photograph dates from 1990, so the location is only approximate.)

	On May 22, 2001, a specially built crate containing the 23 new fragments arrived by Boeing 747 at San Francisco Airport. After being escorted through customs, it was convoyed down to Stanford by a company that specializes in moving cultural artifacts. Where to store such a precious cargo? My office, of course! (It was rekeyed for the occasion.) Watching over the crate from the top of my display case is the left eye of Michelangelo's David, a momento of the Digital Michelangelo Project.
	Note the seal on the side of the crate in the previous image. It's the logo of Lloyd's of London - the same people who insure steamships, satellites, and Springsteen's voice. Note also the two padlocks on the lid of the crate. The crate was accompanied from Italy by Laura Ferrea - an archaeologist with the Capitoline Museums in Rome, who carried the only key.
	The next morning, we opened the crate. It's a marvel of Italian design, consisting of several hand-crafted wooden cases, padded within and without, and fitted to slide into the crate with nary a millimeter to spare. Each fragment was individually wrapped in tissue paper and nestled in its own custom-cut bed of foam. In this image, PhD student James Davis (at left) unwraps the first fragment, while Marc Levoy and Laura Ferrea share a light moment. (Photo by Linda Cicero of the Stanford News Service.)
	Things get serious when we start handling the fragments: two hands at all times, eyes focused on the artifact. By wearing white gloves, we minimized the transfer of oils from our hands to the marble surface. (Photo by Linda Cicero of the Stanford News Service.)
	Fragments larger than 20 cm in length (about half of them) were scanned on the Stanford Large Statue Scanner, the same one we used to scan Michelangelo's statue of David. At right is a closeup of a scan in progress. The laser stripe sweeps across the fragment surface at 1 cm per second. By analyzing the changing shape of the stripe as seen from the side by a precision video camera, the hardware can estimate the three-dimensional shape of the surface, at least along the path of the stripe. The resolution of the 3D data is 1/4 mm.
	Smaller fragments were scanned on the Cyberware Model 15 scanner. To capture all sides of the fragment, while at the same time keeping the laser roughly perpendicular to the fragment surface (to maximize quality), we had to turn the fragment about 20 times, sometimes propping it at odd angles. The resulting 20 range images will be aligned and merged offline to produce a single 3D model for the fragment.
	As we did in Rome two years ago, we separately photographed each fragment using a color digital still camera. Although the two floodlights facing the camera seem bright, the small but focused spotlight standing between the camera and the fragment is actually brighter. It is also placed so as to cast a raking light across the fragment surface, a lighting configuration frequently employed when photographing incised stone artifacts.
	Here are photographs of some of the fragments, shown at the same scale. (The largest fragment is 40 cm long.) The leftmost fragment (fn8) seems straightforward: rows of shops lining a network of streets. The "V" symbol near the bottom of the photograph is one of the plan's standard symbols for a staircase. The middle fragment (fn44) shows similar rows of rooms, but the spaces they open onto are walled off from one another, so they are probably not streets. There was also much discussion about the rectangle of columns (or trees) at the top. The rightmost fragment (fn9) is one of the most remarkable on the plan. It shows shops lining a street, but they are separated from the street by what appears to be covered arcades. Note also that the street is painted red. This is the only known example of a colored street on the plan, and it poses new questions about the purpose of the plan. (See the addendum below for a proposed placement of fragment fn9.)
	After three days of hard work, the fragments have been scanned and photographed. They now lie safely repacked in their crate, awaiting transport back to Italy. As if this week weren't busy enough, we had also just finished carving and breaking the Forma Aedificii Gatesensis. So on Friday evening, the team retired to Evvia in downtown Palo Alto for dinner. Clockwise from lower-left is Marc Levoy, Laura Ferrea, Natasha Gelfand, Steve Marschner, Jennifer Trimble, Nicolas Scapel, Sean Anderson, David Koller, Szymon Rusinkiewicz, and Tina Najbjerg.

23 newly discovered fragments of the Forma Urbis Romae (and our first match)

Addendum - our first match

23 newly discovered fragments of the Forma Urbis Romae
(and our first match)