The Greensboro News & Record has North Carolina dairies crying over spilt milk.

The Greensboro News & Record uncovered a major bid-rigging scheme where North Carolina dairies defrauded school systems of approximately $8 million between 1982 and 1989 by manipulating milk contract bids, requiring the newspaper to manually collect and compare data from 133 separate school districts.

Federal and state investigators believe that between 1982 and 1989, North Carolina dairies rigged bids for school milk contracts and got as much as $8 million of illicit revenue, 2 cents at a time, from the state's school children.

The News & Record first considered looking into the bidding in July 1991, just after we got our newsroom computer and about two months after National Public Radio reported that a North Carolina-based dairy had been implicated in school milk bid-rigging in Georgia.

At first, we had very little to go on. We assumed that the U.S. Department of Agriculture or the state Department of Public Instruction would have bidding information for school milk contracts in a computer file in one central location. They didn't. The records were kept on paper at 133 separate school system offices.

But in August, after another dairy admitted having rigged bids in a school system in an adjacent county, we jumped in anyway. We requested (via form letter) information from all 133 systems, back to 1981, on which dairies had bid on a contract, what kind of contract it was (whether it kept a price firm for the entire school year or allowed the price to vary with changes in the cost of raw milk, which was set by the state dairy commission), and how much per half-pint of milk they were charging.

What we got back created a stack of paper 2 feet high. The only help our tape drive offered was to give me a place on which to set the stack.

Because schools were required to keep such records only for three years and three months, many systems had records only as far back as 1988. And, with school about to start, many child nutrition directors felt like they had more important things to do than help us.

But other systems fell all over themselves to give us everything they had. And some of them had a lot: About 40 systems had usable records all the way back to 1981.

We also got an unexpected break. Early on, one school system's child nutrition director told us that federal marshals had told her that her system definitely had not been touched by the bid-rigging. We were able to use bids in that system as a benchmark for comparing bids in other school districts; in almost every case, the differences were dramatic.

From mid-August until Christmas, I divided my time between typing in the information we had received and calling school systems to beg, plead, cajole, and just generally make enough of a pest of myself to get them to give us what we'd asked for, or fill in the blanks where information we got was incomplete.

Comparing bid prices statewide with bid prices in the system where no bids had been rigged lent credence to investigators' claims that the dairies had received about $8 million in illicit revenue from milk sales between 1982 and 1989.

It wouldn't have been cost-effective for us to sue the holdouts, and ultimately we got at least some usable information from 112 systems.

Meanwhile, I ran up a huge bill in our library, using VuText, DataTimes and other databases to check publications from New Jersey to Texas for valuable information on school milk bid-rigging in other states.

I talked to dairy industry experts who told us what kind of bidding picture we could expect if bidding had been honest-and what we could expect if it hadn't. Then, during Christmas week, I sat down to crunch the numbers.

The analyses and other information showed that:

North Carolina dairies frequently enjoyed exclusive, long-term relationships with school systems in the towns or counties where they were located.

In any given year, prices bid by a particular dairy generally varied by 15 percent or more. One dairy's bid prices varied 60 percent. Our experts considered those swings significant in an industry in which, because of perishability, profit margins seldom exceed 2 percent.

Those swings also were consistent with investigators' suspicions that dairies had bid intentionally high prices in systems where they had no intention of competing for the contract. Such bidding created the appearance of competitive bidding where there really was none.

I also traced the movement of milk prices against the dairy commission's changes in the cost of raw milk that the processors had to pay. (The commission had been disbanded in 1990 and its records destroyed, but a source in state government was able to fax me copies of commission records dating to 1981.) I found that the prices bid by dairies tended to go up more quickly, and some came down more slowly, than the cost of raw milk, the dairies' biggest expense. That meant, if all other expenses were equal, the dairies were making more money. In fact, between 1982 and 1986, their second biggest expense, the price of gasoline, was falling, meaning even more profitability.

The analyses did not conclusively prove bid-rigging. In fact, despite subpoenas for dairy company travel, expense and telephone records, the feds couldn't prove bid-rigging here until two dairy executives secretly began cooperating with them in the spring of 1991. But in records from 1982 to 1989, when word of the federal investigation in other states spread to North Carolina, the analyses did show a strong similarity between the bidding patterns of the three dairies that had admitted rigging bids in North Carolina and the four other large dairies that had not. (Three of those have been implicated in other states.)

This was a highly technical subject. A spreadsheet program would have helped tremendously during the analyses and in generating graphics that would have made the stories clearer to the reader, but we don't yet have one.

There weren't a lot of real people in the package with anything meaningful to say. The only helpful source I could find in the state's dairy industry stayed strictly off the record. And it's not at all clear yet what impact the package has had. Since publication on March 1, seven months after our investigation began, no new indictments have been handed up or guilty pleas announced. The investigation continues at this writing. The project gave me some practice in setting up relational databases working from paper records, and it helped demonstrate to other staffers the value of computer-assisted reporting.