From the introduction:
When it comes to legal method, Holmes is well known for two claims: first, that lawyers and legal scholars should employ an empirical rather than a deductive method and second, that the task of lawyers is to predict the actions of judges. These claims raise a number of questions. Three questions, in particular, seem to recur among thoughtful readers: (1) Why does Holmes, leader of the revolt against formalism, place so much emphasis on the role of logical analysis in law? (2) If legal method is empirical, what is law "empirical" about? What are the facts or the data upon which an empirical science of law should be based? and (3) Under the predictive theory of law, what is it that a judge is supposed to do when (s)he decides a case? Does the judge decide a case by predicting his or her own decision?
Catharine P. Wells. "Holmes on Legal Method: The Predictive Theory of Law as an Instance of Scientific Method." Southern Illinois University Law Journal 18, (1994): 329-345.