Three more bodies and the cockpit shell were recovered from the wreckage of the Air Florida jetliner in the Potomac yesterday, but divers were unable to bring up the two recording devices that could help reveal the cause of the crash.

According to Navy Lt. Cmdr. Stephen Delaplane, who is directing operations at the scene, the flight data recorder and the cockpit voice recorder still are emitting the electronic pings that help divers locate them. But the devices are believed to be buried beneath wreckage and mud on the river bottom.

The operation's risk to the divers became apparent yesterday when a jagged 30-foot section of fuselage severed a canvas sling as it was being hoisted from the river. The section, which contained at least three bodies, plunged back to the river bottom.

In recovering the exterior shell of the cockpit, divers found that the flight deck and instrument panel had been torn away.

The body of the pilot, Larry Wheaton, was found on the river bottom Monday and identified yesterday. National Transportation Safety Board member Francis H. McAdams told reporters last night that Wheaton died of "traumatic injuries." A passenger, whose body was recovered Monday, was identified yesterday as Tom Fisher of Frederick County, Md.

D.C. police said the three persons recovered yesterday were identified as Eric Kauffman of Walkersville, Md., Robert P. Shubinski of Annandale and Spencer Hazelwood. Hazelwood has not been previously identified as a passenger on the plane. According to an Air Florida spokeswoman, he was traveling on a ticket issued in the name of T. Pibbs and was initially identified that way by the airline.

To date the bodies of 56 passengers and crew have been recovered and identified. A total of 78 persons--74 from the plane and four persons on the 14th Street bridge--were killed in the crash of the Boeing 737 last Wednesday.

Four survivors from the plane remained hospitalized yesterday, three of them in "very good" condition at the National Hospital for Orthopaedics and Rehabilitation. The fourth, Patricia Felch, was visited by First Lady Nancy Reagan at Washington Hospital Center where her condition was termed "fair and stable."

Jeanette Bigelow, who was injured when the plane struck the 14th Street bridge, was released yesterday from National Orthopaedics.

Delaplane repeated yesterday that there was "no absolute guarantee that we will recover all of the bodies or every bit of the airplane." He said, however, that the combined military and civilian forces working on the recovery were doing their utmost to reclaim everything possible from the river bottom.

The only portion of the plane brought up relatively intact was the galley.

Hoisted onto a steel barge and opened, the galley still contained closed containers of food, cutlery and other serving pieces for the meal that was never served.

Along with the other recovered wreckage, the galley and cockpit will be examined at Hangar 12 at Washington National Airport.

The structures group there has been examining the left wing, the tail assembly and other pieces of debris that have been recovered for damage and taking measurements on the moving parts, such as flaps.

The systems group also was there to examine hydraulic and electrical systems in the recovered wreckage.

The air traffic control group, working at a nearby hotel and federal offices, continues to examine partial transcripts that they have so far received of eight radio frequencies.

The witness group is interviewing witnesses at the Twin City Marriott. About 100 witnesses have been interviewed.

The overall inquiry involves 70 full-time investigators from the National Transportation Safety Board, Air Florida, the Federal Aviation Administration, Boeing and other parties.

Although approximately 35 percent of the aircraft has been recovered, some of the most valuable portions from an investigative standpoint, the recorders and the two jet engines among them, still are on the river bottom.

Flight data recorders and cockpit voice recorders do not necessarily solve the mystery of a crash. In a few cases, pilots' last words have pointed dramatically to the cause. But more typically the mass of raw data the recorders yield takes weeks to analyze and becomes just one piece in the overall puzzle, though an extremely important one.

The voice recorder is fed by a cockpit microphone. On a continuous 30-minute loop of magnetic tape, it records pilots' words as well as other sounds in a cockpit--controls being activated, the whine of engines, and so forth.

The other recorder preserves flight data on a five-inch-wide foil of nickel alloy. As the foil inches by, four styluses scratch out graphs recording air speed, bearing, altitude and "vertical g-forces," forces that bounce a plane up and down. A fifth stylus makes a record of when cockpit microphones are activated, to allow precise synchronization with the cockpit tape, while a sixth point records the passage of time.

If and when the voice tape is recovered from the Potomac, it will be taken to a special laboratory at NTSB headquarters and dried out. Then a special subgroup of the investigating team will listen for clues in the pilot's final conversations, if any, and the background noise.

If the plane stalled, as one theory goes, the standard stall warning, shaking of the control stick might be audible. If power was lost in one engine, that, too, might be heard by investigators.

Also at the lab, the data recorder's foil will be put under a special microscope, with which an operator will transfer the data to a computer section by section.

If the Air Florida plane took off caked with ice that hampered the wings' ability to lift the jetliner, as one theory goes, the flight recorder graphs could show this.

The air speed graph would indicate normal acceleration on the runway but after liftoff the altitude gain would be abnormally slow. Then the g-forces graph would indicate a shudder in the plane's body. Then would come the rapid descent.

McAdams told reporters that the consensus of pilots interviewed by investigators is that the Air Florida plane had ice on its fuselage, but there was no consensus on whether there was ice on the wings. Four pilots reported seeing ice on the fuselage, McAdams said. Although ice on the fuselage would add weight to the plane, it would not change the aerodynamics of the plane as would ice on the wings.

McAdams said that statements suggesting that the runway was slippery have been obtained from two pilots whose planes landed shortly before the Air Florida plane took off. The pilots said their braking action was poor, McAdams said. Automatic tapes broadcasting from the tower at the time were advising pilots that braking action was fair.

The icing theory is a major focus of the investigation, but there are a number of other possibilities, including loss of engine power, and aviation experts caution that the recorder data alone would not prove any one theory. Any theory would have to be supported by such things as the pilot's recorded words, witness and survivor accounts, examination of the wreckage and past test data.

This process leads investigators to emphasize that analysis of the recorder data is an art, not a science.