This is a book about the law's failure as a system of empirical inquiry. While the US Supreme Court repeatedly says that the aim of a trial is to find out the truth about a crime, there is abundant evidence that many of the rules of evidence and legal procedure are not truth-conducive. Quite the contrary; many are truth-thwarting. Relevant evidence of defendant's guilt is often excluded; reasonable inferences from the available evidence are likewise often excluded. When a defendant elects not to testify, jurors are told to draw no inculpatory inferences from the former's refusal to be questioned. If evidence of prior crimes committed by the defendant is admitted (and often it is excluded), jurors are strictly told to use them only for deciding whether the defendant lied during his testimony and not as evidence of his guilt. Making matters worse, the most important evidence rule of all (saying that defendant can be convicted only if there are no reasonable doubts about his guilt) is monumentally vague; and judges are under firm instruction to decline jurors' frequent requests to explain what a 'reasonable doubt' is. Lastly, this book examines the fact that American courts collect little information about how often they convict the innocent and no information about how often they acquit the guilty. This is tragic because ignorance of the error rates in trials and in plea bargains means that citizens have no grounds for confidence in the judicial system; such a condition of non-transparency should be unacceptable in a democracy. Reform is urgent and this book sketches some of the necessary changes.