A federally sponsored study released yesterday found widely different sentencing practices among District of Columbia judges and recommended a number of guidelines to encourage, but not require, greater conformity.

One D.C. Superior Court judge, for example, incarcerated 87 percent of the convicted felons and another judge sent only 29 percent to jail during a comparable period, the study revealed.

Commissioned by the U.S. Law Enforcement Assistance Administration, the study said that variations in sentencing practices were sometimes hard to explain. The researchers attempted to develop a formula for predicting sentences based on the defendant's background, the particular law violated and the judge's past sentencing pattern, but they found the formula worked only about 60 percent of the time, they study said.

"The judges just don't know what the other judges do with similar cases," said William Hamilton, president of the Institute for Law and Social Research (INSLAW), the government-subsidized group that surveyed the sentences of 1,665 adult felons arrested in 1974.

Hamilton said he does not believe that Superior Court sentencing practices have changed substantially in the last four years. U.S. Attorney Earl Silbert said he had not yet read the the study.

The study divided sentences into two broad categories of "in" (incarceration for any period whether in an adult or youth offender facility) and "out" (fine, probation or suspended sentence).

The study found that "out" sentences were given to:

One-fifth of all convicted robbers.

One-third of all convicted burglars.

One-eighth of all convicted rapists.

Half of those who were convicted of ared assaults.

Three-fifths of those convicted of assaulting a police officers.

Hamilton said the INSLAW report should not necessarily be taken as a call for tougher sentences. "But it's clearly wrong that they lack even-handedness," he said. Judges at the opposite ends of the sentencing spectrum "can't both be right," he said.

The study said that some of the differences amont g the 38 judges whose sentencing practices were surveyed could be attributed to variations in the seriousness of the cases that came before them. But judicial philosophy also appeared to be a significant factor, according to Terence Dungworth, the criminologist who directed the study.

Felony sentencing provisions here, which allow judges wide latitude, are scattered throughout the D.C. and U.S. criminal codes, and "have not undergone major revision for more than 30 years," according to the INSLAW study.

Neverthelss, Hamilton said, judges in the middle of the sentencing spectrum appear to have developed some "common sense" rules.

The report suggests that the best way to reduce variations in the sentening of comparable defendants may simply be to provide judges with more information. Under current court procedures, a probation officer prepares a presentence report on each defendant, dealing mostly with the defendant's history, dangerousness and prospects for rehabilitation.

The INSLAW report proposes a new form of presentence report that would include a "historical sentencing profile" of what the court has done with comparable defendants convicted of the same offenses.

One finding that surprised law enforcement officials, according to Hamilton, was the large number of defendants-58 percent-given "out" sentences for the crime of assault on a police officer. Prosecutors told the INSLAW team that only "infinitesimally small" proportions of these cases were prosecuted as such; the rest were generally broken down to simple assault or another lesser offense.