Last week, when the Senate Finance Committee voted subsidies to help poor people pay rising fuel costs, the senatorial talk was all of poverty and freezing temperatures, of abstract concepts like "heating degree days," "cooling degfree days," "pass-through of indirect energy costs."

Every once in a while, a senator would earnestly assure his fellows that all he sought was an equitable formula for distributing the $2 billion or so a year in fuel aid to the poor in 1981 and 1982.

From behind these high-minded as suracnes, however, a more grubby reality peeked out: the formula that seemed fairest and most equitable to each senator usually turned out to be the formula that helped his state to a bigger slice of the $2 billion pie.

So it was perfectly predictable that the committeed -- with a big majority of members from cold, northern states -- quickly junked President Carter's proposed split, which would have given a sizable amout of air to poor people in the South. It was contemptuously dismissed by one northern Democrat as a "florida primary formula."

Instead, the committee adopted an amendment by Gaylord Nelson (D Wis.) that gave a much larger part of the $2 billion to northern states, while cutting the share of every southern state but Virginia, whose senior senator, Harry F. Byrd Jr., just happens to be a committee member.

A Republican substitue would have made off with an even larger share for the northern states, but the committee's northen Democrats, recognizing that their party still has a major southern wing, settled for the Nelson compromise.

In going by the state benefit numbers and the printouts -- the big computer readouts that show exactly how much each state will get under different formulas -- Nelson and his colleagues were playing one of the most common though little-noticed forms of Capitol Hill politics.

It is the formula game -- getting more money for your state or congressional district when it is time to write formulas divvying up the billions of dollars that flow out of Washington in grants to the states and localities.

The stakes are not trivial. This year, the federal government will pump out about $83 billion to states and localities in grants.

The $83 billion, most of which is distributed through such formula programs as Medicaid, welfare, revenue sharing and community and economic development assistance, is about one sixth of the entire federal budget, and makes up over one-fifth of state and local government revenues.

The game's importance is shown by this: when Nelson pushed his formula through the Finance Committee, it boosted Wisconsin's 1981 share of the first $2 billion of fuel aid to the poor from 1.97 percent to 2.74 percent. Instead of $39.4 million, people in his state would get $54.8 million. Alabama, on the other hand, dropped from $39.8 million to $29.2 million.

The formula game, the behind-the-scenes maneuvering and tinkering with apportionments and formulas to pry loose more money for your own region, is played every day.

It has been intensifying as the federal budget tightens, because now there isn't the hope of unlimited new federal spending so that everyone can do well; participants must take from one another. Battles between northern and southern states are becoming so intense that members and staff jocularly talk about the "Second Civil War."

Every formula is complicated; usually it takes a computer to determine how much a state or local government will get. But some principles are well known. If you want to help the South, and particularly the rural and smalltown South, write the formula to give more money to regions having lots of poor people and a low per-capita income. Despite talk of southern gains and northeastern decay, the percentage of people under the poverty line is far higher in the South than the Northeast -- over 15 percent compared to 9 percent -- and is the highest in the country. The Southeast's per-capita income, $6,773 in 1978, is well below that of any other region, including Mid-Atlantic ($8,230) or even "decaying" New England ($7,900).

If you want more money in a formula for the Northeast and Midwest, crank in a factor based on unemployment, which in recent years has been higher in these older industrial areas, or age-of-housing (the North has more houses built before 1940). Or go by the number of people on the state welfare rolls and how much the state spends for them. Although the North has fewer poor people, the big industrial states usually have easier eligibility and higher benefits. So their welfare rolls and costs are much higher.

Some of the most ferocious legislative battles of recent years have involved formulas.

The dispute over revenue sharing in 1972 is a good example; formulas were changed at every twist and turn to benefit a party or region.

The basic ida was for the federal government to hand over a certain amount of money to the states and local governments each year -- $5.3 billion the first year -- to be used pretty much as they wanted.

"When Richard Nixon first sent it up," a veteran of those struggles recalls now, "he showed a distribution formula in a brown book which was basically a suburban formula. It gave lots of money to the wealthy suburbs where the Republican power was."

Sen. Edmund S. Muskie of Maine, then the leading contender for the Democratic presidental nomination immediately denounced the proposal as making the "rich richer." But he soon decided that he soon decided that he should get firmly behingd the popular notion, though with a different formula.

Muskie quickly drafted an alternative formula that depended in large measure on the number of welfare clients and poor people in an area. The welfare-client factor in this formula gave lots to the big cities of the North; the poor-people factor gave lots to the southern cities and rural areas too.

"It caught the cities whether North or South," and thus was a sound formula for a Democratic candidate.

The action then moved over to the House Ways and Means Committee, where Chairman Wilbur Mills (D-Ark.) at first had opposed the revenue-sharing idea but finally decided to go along.

Mike Bird, and economist for the Joint committee on Taxation, remember, "We drafted an alternative for Mills to distribute the money primarily on the basis of the number of poor people." This was most favorable to the South and rural areas, and, by a quirk which Bird insists wasn't deliberate, the very first version of the Mills formula Quickly changed to avoid embarrassment) have Little Rock the highest per-capita revenue-sharing allotment in the country.

Other Ways and Means members, many from northern and urban areas, took a look at the computer printouts for their regions and didn't like what they saw. So says bird, another formula with five major factors was prepared. It gave weight to urbanized population, state individual income taxes collected and general tax effort, all of which helped the urban areas of the North and Northeast

This passed the House, but in the Senate, which is generally more rural and southern in makeup, "they didn't like the House formula. It didn't give enough to the South and rural areas and the West."

So the senate dropped from the formula two to the factors that favored richer areas and northern cities. As a result, very poor areas (read South and rural) did somewhat better.

In the House-Senate conference there was an impasse. So in its wisdom, Congress decided to let each state and community take either the House or Senate formula.

Money comparisons of the House version (heavy urban tilt) and Senate version (tilted more rural and southern) are illustrative. Mississippi, poor, southern and rural, got $45.7 million under the House formula but $97 million under the Senate. New York, well-to-do, northern and urban, got $643 million under the House formula and $501 million under the Senate.

A well-publicized Sun Belt/Frost Belt clash occurred in 1977 when the Community Development Block grants program (providing $3.2 billion this year) came up for renewal. The dispute centered on how heavily the age of housing in an area should weigh in the distribution formula. A southern-backed attempt to reduce the weight given to pre-1940 housing was beaten on the House floor.

In recent months, formulas for medicaid and welfare have received considerable attention from Sen. Jacob K. Javits (R-N.Y.) and Rep. Charles B. Rangel (D-N.Y.) of Harlem.

medicaid is the biggest federal grant program, pumping out over $12 billion a year to the states to help them pay for free or low-cost medical treatment of the poor. Welfare (AFDC) is another big one -- $6.7 billion.

Both reimbursement formulas (they use the same one) favor poor areas, on grounds that poor states are least able to afford charitable outlays. States with lowest per-capita income get the highest reimbursement. New York, New Jersey and other urban industrial states get back 50 percent of each dollar they lay out, while poor states like Mississippi get 78 percent.

Recently, Javits and Rangel took up a concept hatched several years ago by the Advisory Committee on Intergovernmental Relations: an index of "tax capacity" for each state. It takes into account not only per-captia income but such other wealth or income as severance taxes, mineral resources and business profits, New york would get back a lot more than 50 percent if a pure "tax capacity" formula were used -- exactly what Javits and Rangel had in mind.

However, the calculations showed that 25 other states' percentages would go down. Since this was politically impossible, they decided to put in a 55 percent reimbursement minimum so that nearly everyone would benefit. New York would go up to 59.2 percent. then, just to make sure of political salability, they wrote in a provision saying that if you got more under the old formula, use that.

But it would cost the United States a lot more, and this isn't an era of bigger outlays. Right now it's waiting in the wings.