They are Uncle Sam's all-purpose appraisers, the bane of the too-crafty corporation and the me-minded millionaire.

Their job is to catch cheats, as well as innocent mistakes. Give them a charitable contribution, and they will decide whether its value has been inflated for the sake of a bigger deduction. Give them a piece of an estate, and they will weigh whether its worth has been slighted to cut inheritance taxes.

They are technical employes of the Internal Revenue Service's engineering and valuation branch, and they seem harmless enough, beneath five layers of IRS bureaucracy. But they are in the business of appraising property -- valuable property -- and their judgment can move millions.

The 58 IRS district offices turn to them for help when an audit bumps into an item of uncertain value. The valuation staff tries to give it a price tag, assess its market value.

It is often a tricky business. The office deals with everything you could lay your hands on -- and some things you can't -- from paintings to patents, skyscrapers to swamps. And to make matters worse, the top tool of the appraiser -- looking at "comparables" -- is frequently useless, since many of the items are one of a kind. How, for example, do you affix a value to a patented idea -- by definition unique -- which no one's tried yet to market?

"You make an educated guess," explains Geoffrey Taylor, chief of the engineering and valuation branch. "It's definitely an art and not a science."

And high-stakes art, at that. Taylor and his staff of 64 deal only with items worth many thousands of dollars -- at least by the taxpayer's estimate.

Those facing estate taxes can profit by underestimating the value of the taxable assets, while individuals and corporations making gifts to charity may overestimate their value to claim a larger deduction.

Taylor has an eclectic group to meet the eclectic task before him. A certified public accountant and mechanical engineer himself, his staff of 64 includes foresters, geologists, real estate experts, lawyers and financial analysts. On the appraisal side -- Taylor's staff also provides interpretations of IRS rules -- are groups to deal with equipment and patent work, securities, timber and real estate, and art and historic documents.

Easiest are cases where the appraisers can compare an item with others, recently sold, which are similar -- matching antiques, for example. Having no such comparables, the appraiser will try to evaluate how much income something produces. A final indication of market value is the cost of reproduction. When all three kinds of evidence provide little help, it is time to make an educated guess.

"The degree of precision varies, but the law requires a determination," says Taylor. "We try not to be arbitrary."

A special difficulty arises where there is a "conceptual" disagreement about how a piece of property can be used. Take an intersection of two major highways, Taylor suggested.

"There's a situation that could make an ideal location for a hotel. The guy there may be growing potatoes today, so obviously the price he'd get for the potatoes is no indication of what the property's worth."

Because the craft is imprecise, Taylor's staff changes appraisals only when it finds the taxpayer's figure far off base.

"When you look at the type of property, there's a certain band of reasonableness," he says. "If [the taxpayer] is within that area, we don't challenge it."

With a large estate, a regional IRS office will ask Taylor to appraise only a few items. If the taxpayer's figures for them prove reasonable, all of the appraisals are accepted.

Taylor occasionally hires outside consultants but the IRS has a standing art-advisory panel of dealers, museum curators and university experts that places price tags on all kinds of art objects. They serve the agency for the modest sum of plane fare plus $50 per day.

"Too many times things are done by amateurs, bureaucrats who don't know what they're doing," says Lawrence Fleischman, a panel member and partner of the Kennedy Galleries in New York. "It's important that things be kept straight. I believe in American art, and I don't like ripoffs. I don't like people who try to get away with something."

"There are many that are flagrantly dishonest," Fleischman says. "Somebody will put an amateur's work in, and put some gigantic figure on it. They try to palm it off on some unsuspecting philanthropic organization."

The panel reviewed 440 items last year, and changed the value on 278 -- 63 percent of them. Taxpayers claimed they were worth almost $31 million, and the panel made more than $6.1 million in adjustments.

Even the most flagrant of cheats will rarely face prosecution, says Taylor, because it is so difficult to prove willful intent to deceive. A taxpayer can easily blame his inaccurate figures on his private appraiser. And few tax returns -- 2 percent for individuals, less than 8 per cent for corporations -- are ever audited in the first place.

Taylor's staff usually has no contact with the taxpayer except when they need to inspect property or equipment. Instead, the district IRS offices deal with the "customers" in the 600-odd cases the engineering and valuation staff handles each year.

Taylor says he has trouble relating "on a personal level" to the sums he deals with. But he believes the system "must be working" since more taxpayers aren't appealing his staff's rulings.

Now 54, Taylor reached his post during two decades at the IRS, climbing first to group chief, then to section chief, and finally, in 1977, to branch chief. Like many bureaucrats, he is comfortable where he is, and expects to remain. "The only plans I have," he says, "are to stay here."