The Arlington School Board apparently has narrowed to two the grade grouping formats it favors using when county schools are consolidated in 1984.

Although the board could make a final decision on grade formats at its Oct. 21 meeting, it is expected to ask Superintendent Charles E. Nunley and his staff to study the favored formats and to report their conclusions at a December board meeting. The study would include how the plans could be implemented and at what costs.

At the board's meeting last week, three of the five members said they would favor changing the current system only if an alternative is shown to be, as member Michael E. Brunner said, "demonstrably better."

The current grade format groups students in kindergarten through sixth grade in elementary schools, grades seven and eight in intermediate schools and grades nine through 12 in high schools.

If the board continues to use this format, one of the county's three regular high schools probably will have to be closed if enrollments continue to decline. Closing one of the schools is expected to create bitterness in the community served by the school.

Still, said board member Simone J. (Sim) Pace, the format "works and we know how to do it." Board Chairman Evelyn Reid Syphax agreed with Brunner and Pace.

The other option, which appeared to be favored by members Margaret A. Bocek and Claude M. Hilton, would keep the county's three high schools open but would eliminate intermediate schools. Their proposed format would group students in kindergarten through the seventh grade at elementary or intermediate school buildings and eighth through 12th grades at the high schools.

None of the members gave a high priority to a format advanced by Washington-Lee High School teachers, parents and students. That proposal would keep the current format for elementary and intermediate schools, but would send ninth and 10th graders to two high schools and 11th and 12th graders to a third "super" high school.

Still, some of the board members said the Washington-Lee plan should not be completely ruled out, and Brunner and Pace gave it lukewarm support.