For the past nine years, two Washington elementary schools at the opposite poles of educational reform have been part of a well-financed national experiment trying to find out how poor children might learn more.

At the Nichols Avenue School in Anacostia the teachers have been going by the book - a series of highly detailed workbooks, called the Distar instructional system, that are accompanied by scripts for the teacher for each day's lessons.

At the Morgan School, 18th and California Streets NW, the emphasis is on freedom and choice for both teachers and students. Reading and mathematics are taught in a variety of ways, mostly in informal open classrooms.

Although children at the two schools are from similar low-income backgrounds, their levels of achievement, according to standardized tests, are strikingly different. Those at Nichols Avenue score much higher than those at Morgan.

The same pattern is repeated throughout the country, according to a major new evaluation of the federally-financed Follow Through program, in which both Morgan and Nichols Avenue participate along with schools in about 100 other cities.

The study, by Abt Associates of Cambridge, Mass., reports that highly structured programs that emphasize basic skills have been much more successful than open classrooms in raising the achievement of low-income children.

In fact, the study says, youngsters in the open classrooms, even though they had extra teacher aides and expensive innovative equipment, actually achieved less in reading and mathematics than similar children in regular classrooms.

The structured programs also succeeded in raising the pride, self-condidence and sense of responsibility of the low-income children who took part in them, the stusy says. On the other hand, it says, the open classroom progress generally failed to raise self-esteem even though most sponsors of these programs said they try to do that first with the expectation that children who feel better about themselves will learn more.

Instead, the report says, the process apparently works the other way around with children who have academic success then feeling better about themselves.

The new $3 million study, which was financed by the U.S. Office of Education, uses data on about 9,200 third-graders in all parts of the country who had been in Fellow Through for at least three years. Comparison groups of low-income children were also tested in regular classes.

Since Follow Through started in 1968, the federal government has spent about $500 to serve about 80,000 children in kindergarten through third grade.

The program was orginally proposed by President Johnson as a large-scale follow-up to the federally financed Head Start nurseries. After Congress appropriated less than Johnson requested, it was set up as a "planned variation" research experiment with about 20 different colleges and companies given money to sponsor, model programs to test a broad range of educational philosophies and techniques.

According to the new report, of all the programs tried, the most successful was Distar, which was sponsored by the University of Oregon.

Overall, third-graders who had been taught by Distar since kindergarten had reading scores at the 41st percentage for al children of 50), and had mathematcs scores at the 48 percentile.

At Nichols Avenue the third-grade reading scores last year were at the l35th percentile while mathematics was at the 56th percentile.

The only other major Follow Through model to average above the 30th percentile in either reading or math was Behavior Analysis, another highly structured program that was sponsored by the University of Kansas.

In comparisons with similar children not in Follow Through, the only program besides Distar to show generally positive effects was the bilingual language development approach, sponsored by the Southwest Educational Development Laboratory (SEDL) which also fell into the structured category.

The Abt report said that overall the children in Follow Through did worse in reading and mathematics than those in regular classrooms, probably because many of the innovative programs stressed "learning to learn" and "tedned to divert the child's attention from the basics."

The program at Morgan School is part of the open education model, sponsored by Education Development Center of Cambridge, Mass.Overall, the EDC third-graders had reading scores at the 18th percentile. At Morgan the results in both reading and mathematics were at the 12th percentile.

Richard B. Anderson, director of the evaluation for Abt, cautioned that there had been serious problems in collecting and interpreting the data. He noted that although the averag, the structured programs had been much more successful than non-structured ones some of the structured programs worked much better than others. Even Distar, which was the most successful program, did not do well everywhere it was tried, he said.

Although the programs can be ranked by their average results. Anderson noted that every one of them had a wide reange of success, indiecating that particular teachers and schools do have a powerful effect on how well children learn, no matter what books and techniques are used.

At the Nichols Avenue School the Distar books are used in reading, language skills and mathematics. The lessons for each day are in sequence, numbered from 1 to 170, and teachers (or aides use what is called a presentation book containing a script with directions in black type and what the teacher should say in red. Everything is taught step by step to groups of six or eight youngsters at a time.

There are plenty of test and much repetition but children can move ahead quickly if they've mastered the material. There is homework every night, even for kindergarteners.

Despite charges by critics that Distar is too regimented, Martin Patrick, who teaches kindergarten at Nichols Avenue, said, "I love it. I think teachers can use their creativity the rest of the day, not in teaching reading and math.Once the children find out they can read, they get excited about it. When they can do (math) problems, it makes them feel big and they show off."

James Guines, the District's associate school supereintendent for instruction, said Distar also is used in some prekindergarten programs and in a smattering of other elementary classrooms around the city. He said, however, that there was no complete listing of where it is in use.

"We have about 30 different reading programs being used in D.C. public schools," Guines said. "We leave the decision-making up to the teacher. In most school systems there are just a few adopted textbooks that everybody must use, but we're very democratic."

Guines said the D.C. school system has not been able to compare how well different programs do because for the past four years standardized tests have not been given in every school.

He said that next year the standardized tests in reading and mathematics will be taken by all students in the third, sixth, and ninth grades. In addition, Guines said teachers will be given check-lists next fall of reading and math skills that children are expected to learn in each grade. He said teachers still will be able to use any method or textbooks they prefer.