Over three weeks in June 2005, the Court of Appeal in London heard four appeals against convictions for non-accidental injury of infants: one murder, two manslaughter, and one grievous bodily harm. The three appeal judges delivered their judgement in a 67 page document on 21 July.¹ Two appeals were upheld on the grounds of process, not medical evidence, one was dismissed, and in the fourth the conviction was reduced from murder to manslaughter.

The basis of the appeals was that since these convictions in the 1990s and 2000, new research had suggested that the long held belief that infants who presented with encephalopathy, thin subdural haemorrhages, and retinal haemorrhages—the triad indicating “shaken baby syndrome”—had been subjected to extreme and repeated violence was wrong and that little or no trauma need be involved.

The research in question has become variously known since its publication in 2003 as the “Geddes hypothesis” or the “unified hypothesis”.² It was based on a pathological study of the dura of 50 intra-uterine, neonatal, or infant deaths, which identified microscopic haemorrhage within the layers of the dura in 36. This led to the speculation that subdural and retinal haemorrhage was not caused by traumatic shearing of subdural veins and retinal vessels but by a combination of cerebral hypoxia, raised intra-cranial pressure from brain swelling, raised arterial pressure, and raised central venous pressure. The publication of this research was met with considerable scepticism by most working in the field of paediatrics, paediatric pathology, and paediatric head injury,^3–7 but it was enthusiastically embraced by a few^8, …