Hung juries and majority verdicts: what happens when twelve people cannot agree

English criminal justice tells two stories about the jury at once. The first is that guilt must be proved beyond reasonable doubt to twelve ordinary citizens. The second, quieter story is that the system will not actually wait for all twelve, and that when they cannot agree at all, the machinery for deciding what happens next is governed less by principle than by pragmatism, cost and a prosecutor's judgement about whether to try again.

Unanimity was the rule in England and Wales for more than six centuries until the Criminal Justice Act 1967 introduced majority verdicts. The stated justification was jury "nobbling": a bribed or intimidated juror could no longer single-handedly force a deadlock. The modern rules sit in section 17 of the Juries Act 1974. A full jury of twelve may convict or acquit by 11-1 or 10-2; if jurors have been discharged through illness or misconduct, a jury of eleven needs ten to agree and a jury of ten needs nine. Nothing smaller than 9-1 will do, and a jury reduced below nine cannot return a verdict at all.

The timing matters as much as the numbers. A judge cannot invite a majority verdict until the jury has deliberated for at least two hours, a floor lifted to two hours and ten minutes by practice direction to cover the walk between courtroom and jury room. In long or complex trials judges routinely wait far longer, often days, before giving the majority direction, because an early direction signals that unanimity is negotiable. When a guilty verdict is returned by majority, the foreman must state in open court how many agreed and how many dissented, so the split of 10-2 becomes part of the public record; an acquittal by majority is announced without the numbers, to avoid tainting the defendant with the arithmetic of doubt.

If even ten cannot agree, the judge has one more instrument before conceding failure. The Watson direction, settled by the Court of Appeal in R v Watson in 1988, tells jurors that each must be true to their oath and their conscience, but that a verdict requires argument, listening and necessarily some give and take. Its careful wording is the residue of decades of appellate policing: earlier formulations that mentioned the expense of a retrial or leaned on holdout jurors were condemned as coercion, and convictions extracted under them were quashed. A Watson direction may only be given after the majority direction, and many judges never give it at all, preferring to let a genuine deadlock be a deadlock.

The retrial calculus

A hung jury produces no verdict, and therefore no acquittal. The rule against double jeopardy does not bite, and the Crown Prosecution Service must decide afresh whether prosecution still passes its two-stage test of evidential sufficiency and public interest. In practice the first hung jury usually leads to a retrial before a new jury, months later, with witnesses recalled and costs run twice. If the second jury also hangs, a convention as old as the majority verdict itself takes over: the prosecution ordinarily offers no evidence and the judge enters a not-guilty verdict. The convention is not law, and the Court of Appeal confirmed in R v Bell in 2010 that a third trial is permissible where the case is grave and the circumstances exceptional, but such trials remain rare and controversial.

Scotland runs the experiment differently, which shows how contingent the English settlement is. Its criminal juries have long had fifteen members deciding by simple majority, with the extra option of not proven, and legislation passed at Holyrood in 2025 restructures that model rather than importing unanimity.

Certainty traded for finality

Each stage of this apparatus buys finality by shaving certainty. The 1967 Act decided that two dissenting jurors are noise rather than reasonable doubt. The Watson direction nudges a stalled minority towards accommodation while insisting, somewhat delicately, that nobody is being pressured. The two-trials convention accepts that if twenty-four jurors split without agreeing, the state should stop asking. None of these thresholds is derived from any theory of proof; the number ten, the two hours, the two attempts are administrative settlements, arrived at because trials are expensive and witnesses cannot be summoned forever. A defendant convicted 10-2 has been found guilty by a tribunal in which one juror in six thought the case unproved. The system's answer is not that those two jurors were wrong, but that at some point a verdict must be delivered, and it has quietly decided where that point lies.