physical deficiencies of the first kind on student performance is difficult to measure. The Appellate Division took note of this concession, dismissed as "anecdotal" plaintiffs' evidence of "leaky roofs, deficient heating and other problems," and credited testimony that "all immediately hazardous conditions had been eliminated" (295 2 at 10). Eliminating immediate hazards is not the same as creating an environment conducive to learning, and the record contains much evidence about deficient school infrastructure. Nevertheless, on this record it cannot be said that plaintiffs have proved a measurable correlation between building disrepair and student performance, in general.[4] On the other hand, plaintiffs presented measurable proof, credited by the trial court, that New York City schools have excessive class sizes, and that class size affects learning. Even in the earliest years -- from kindergarten through third grade -over half of New York City schoolchildren are in classes of 26 or more, and tens of thousands are in classes of over 30. As the trial court noted, federal and state programs seek to promote classes of 20 or fewer, particularly in the earliest years, and plaintiffs' experts testified on the advantage of smaller classes. As the 1999 655 Report shows, New York City elementary school classes average five more pupils than those of other schools statewide excluding Buffalo, Rochester, Syracuse and Yonkers. Although the Appellate Division found "no indication that students cannot learn in classes consisting of more than 20 students" (295 2 at 11), plaintiffs' burden was not to prove that some specific number is the maximum class size beyond which children "cannot learn." It is difficult to imagine what evidence could ever meet a burden so formulated; nothing in CFE required plaintiffs to do so. Rather, plaintiffs alleged "fact- based inadequacies" in educational inputs, and we held that the State's failure to provide the opportunity to obtain "fundamental skills" would constitute a violation of the Education Article (86 2 at 319). Accordingly, plaintiffs had to show that insufficient funding led to inadequate inputs which led to unsatisfactory results. Plaintiffs' education evaluation statistics expert Dr. Jeremy Finn showed -- on the basis of the Tennessee Student Teacher Achievement Ratio ("STAR") project and related research - - that, holding other variables constant, smaller class sizes in the earliest grades correlate with better test results during those years and afterwards (187 Misc 2d at 52-53). The trial court found that the State's expert Dr. Hanushek failed to rebut these conclusions, and the Appellate Division, mistakenly addressing a nonexistent claim