In fall 2003, officials of Freedom Area School District brought together 23 low-scoring math students and delivered a bombshell: They'd no longer be allowed to fail.

The district assigned the students, all sixth-graders, to a special class and provided them two teachers, a modified curriculum and an "A, B, Not Yet" grading system meant to reinforce the new way of thinking.

The group's test scores soared.

Freedom Superintendent Ron Sofo recounts the experiment, which fueled other improvement efforts at the Beaver County district, in this summer's edition of the Harvard Educational Review.

Dr. Sofo hails the "Target Math" program as the kind of innovative, grass-roots initiative that's essential if schools are to improve instruction and prepare all students -- even stragglers -- for 21st-century jobs.

He said such efforts also help meet the accountability requirements of the federal No Child Left Behind Act, a law that he criticizes for doing little to foster achievement.

"While NCLB has acted as a catalyst for accountability, it ignores the ways in which public school operations need to be overhauled to promote faculty empowerment," his article said.

Dr. Sofo and then-middle school Principal William Renko first outlined the effort in their 2004 book, "No Bad Schools."

In an interview, Dr. Renko called Freedom's experience a "parable" for schools and policy-makers.

The 23 students had failed math courses repeatedly, and all scored basic or below on their fifth-grade Pennsylvania System of School Assessment math test. None was eligible for special-education services.

In his article, Dr. Sofo said the experiment was designed to make learning more meaningful, yet more rigorous, than it previously had been. He said organizers hoped to build rapport between teachers and students, partly through individualized instruction, and set up a system, including student feedback, for fine-tuning the program.

The math curriculum was pared so the group could spend more time on the most important concepts, and students had greater access to computer math exercises than peers in regular classes. The grading system was the most unusual feature, Dr. Sofo wrote.

"The Target Math team theorized that giving F's to students who already had a history of failure in math would be counterproductive to the program's goals of improved student effort and mastery of math knowledge and skills. 'A, B, Not Yet' grading meant students kept revising nonproficient work (scored Not Yet) until its quality was high enough to warrant a true A or B.

"In other words," he added, "success was to become the only option."

The grading and other aspects of the program drew criticism from some teachers, who claimed bad students were receiving special treatment.

"This is a very radical grading system. I knew how controversial this was going to be in a public-school setting," Dr. Renko said.

By the last grading period of 2003-04, all 23 students received A's and B's, with the lowest a B-. Three performed so well that they were released to regular classes the following school year, where they "did fine," Dr. Sofo said in an interview.

Of the 17 students who stayed in the program through middle school, 10, or 59 percent, scored proficient or advanced on their eighth-grade math PSSA.

"The remaining seven students each faced extraordinary family and life circumstances outside of school that made giving their best effort every day a tremendous challenge," Dr. Sofo's article said.

Though some modifications have been made -- Dr. Renko's successor gave teachers the discretion to use a regular grading system, for example -- the district still operates a Target Math program in grades six through eight. It also launched Target Reading in middle school and introduced block scheduling, laptop computers and other improvement efforts at the high school.

"Such rethinking must start not with top-down mandates but with teachers and principals, who hold the key to classroom- and school-level innovation," Dr. Sofo's article said.