While the Incan empire left nothing that would be considered writing by today's standards, it did produce knotted strings in various colors and arrangements that have long puzzled historians and anthropologists.

Many of these strings have turned out to be a type of accounting system, but interpreting them has been complex.

Now, Gary Urton and Carrie J. Brezine of Harvard University say they have found a three-knot pattern in some of the strings, called khipu, that they believe identifies them as coming from the city of Puruchuco, about seven miles north of modern Lima, Peru.

Now, Gary Urton and Carrie J. Brezine of Harvard University say they have found a three-knot pattern in some of the strings, called khipu, that they believe identifies them as coming from the city of Puruchuco, about seven miles north of modern Lima, Peru.

They used computers to analyze 21 khipu found at Puruchuco and divided them into three groups based on the knot patterns. Their findings are reported in Friday's issue of the journal Science.

One group seems to be for local use and the other two groups _ each with the three-knot pattern -- may have been used to report local activities to higher authority, or to receive messages from those authorities. Details of the information from the local khipu was coded onto the others intended for travel.

In this case, the researchers believe they have found a place name in the three knots. "If that's the case, we should ideally be able to look around at other khipu and see if we see this arrangement," Urton said.

"We suggest that any khipu moving within the state administrative system having an initial arrangement of three figure-eight knots would have been immediately recognizable to Inca administrators as an account pertaining to the palace of Puruchuco," the researchers said.

"For the first time, really, we can see how information that was of interest to the state was moving up and down in a set of interrelated khipu," Urton said in a telephone interview.

"We assume it has to do with tribute, the business of the state, general census taking or what resources existed or what activities were taking place," he said.

Identifying a place-name, they said, could provide the first foothold for interpreting the knots.

Potentially, Urton said, they might be able to build up an inventory of place names, the first time khipu knots have been directly associated with words rather than numbers.

There are between 650 and 700 khipu in museums, he explained, and about two-thirds of them have the knots organized in a decimal system indicating their use in some sort of accounting.

But the remaining khipu have knots in other patterns, perhaps a form of written language, if the researchers can work it out.

"We think those may be the narrative ones, "Urton said. "The identities attached to those knots may not be numerical. If we can use the numericals to account for objects, that may give us clues to how they were assigning identities to objects," he said, citing such items as llamas, gods, defeated cities and warriors that might have been counted.

If they are able to find such words, then they could look for those words in the narrative khipu.

What is missing is something like the Rosetta stone, which allowed Egyptian hieroglyphics to be deciphered when researchers realized it contained identical text in three languages, two of which could still be understood.

The Inca empire flourished along the western edge of South America in the late 1400s, ending with the arrival of the Spanish in the early 1500s. There are reports of the Inca telling the Spanish conquerors that the khipu told history, good and bad. The Spanish reportedly wrote down some of the Inca stories, but destroyed many of the khipu.

Galen Brokaw, professor of languages at the University at Buffalo, called the paper "exciting," because Urton was able to show a relationship between three levels of khipu.

"Each higher level condenses the more specific and detailed information of the level immediately below it. So, this provides us with an idea about how khipu were used in the Inca administration. To a non-specialist, it may sound like a fairly small discovery, but within the context of khipu studies it is fairly significant," Brokaw said.

Heather Lechtman, a professor of archaeology and ancient technology at Massachusetts Institute of Technology -- after hearing a description of Urton's paper -- said "he is making an interpretation, and I expect that he is not far from the mark."

Neither Brokaw nor Lechtman was part of Urton's research team.

The research was supported by the National Science Foundation, Dumbarton Oaks Foundation, Harvard's Faculty of Arts and Sciences and the John D. and Catherine T. MacArthur Foundation.