By any measure, a courtroom is not a laboratory. And although both claim to seek the truth, the motives of lawyers and scientists hardly coincide.

The situation was summarised well in The Times last year: “Our combative courtrooms are based on partisan conflict with all parties approaching their positions subjectively in order to argue their cases passionately. Participants emphasise the facts that favour their side’s story. In contrast, science must be neutral and objective, while looking at all the facts.”

Forgetting this vital distinction can have terrible consequences. Scientists now insist that multiple cot deaths in the same family do not equate to multiple murder beyond all reasonable doubt. Sensibly, convictions under- pinned by dodgy statistics have been overturned.

Now shaken baby syndrome (SBS) is firmly on medical trial. Its showy arrival on the legal circuit — 200 cases come before British courts each year — contrasts with a tepid, if not frosty, reception in medical circles. Last year, the Attorney- General ordered that every parent convicted of shaking their infant to death should have their case reviewed. And last month, the British Medical Journal went further by questioning whether such a syndrome really exists.

SBS has lain in the literature since at least 1972, and its description is most often credited to John Caffey, a retired American radiologist (although the BMJ published a 1971 article by A. Norman Guthkelch on the whiplash forces that could induce brain bleeding now thought characteristic of SBS). As nobody is going to shake a baby to prove a theory, the idea that vigorous shaking could cause a particular pattern of injury worked its way into the medical consciousness.

Ever since, doubts have flourished because the precise types of injuries — bleeding on the brain and in the eyes — supposedly unique to SBS

are also seen in other circumstances, including accidents. Research suggests that so-called subdural (brain) and retinal bleeding may also appear after an infant has fallen, choked on food, or suffered certain infections.

Neuropathologists such as Jennian Geddes, at the Royal London Hospital, have

long pointed out that brain

scans cannot provide incontrovertible proof of malicious intention. Geddes,

in particular, has shown that, among one sample of babies who were classified as victims of SBS, more than half showed evidence of a lack of oxygen in the blood. This could have caused the brain to swell and cause bleeding as a secondary symptom. Oxygen deprivation can be caused by choking, asphyxia or infection. Geddes also suggested that ignorance, rather than ill temper, can lead to similar injuries.

Geddes, now retired, is one of two co-authors of the BMJ editorial. She points to two important reviews of the literature, each of which aimed to piece together all the evidence for SBS as a single syndrome that can be diagnosed without ambiguity. Those reviews are worrying. One suggests that the use of retinal haemorrhage to diagnose shaking as the cause of injury “cannot be supported by objective scientific evidence”; another, spanning studies published over 32 years, found that there was “inadequate scientific evidence to come to a firm conclusion on most aspects of causation, diagnosis, treatment, or any other matters”.

Geddes concludes that, for now, the only correct course of action is for experts to declare their uncertainty in the courtroom. She writes: “Doing so may make prosecution more difficult but a natural desire to protect children should not lead anyone to proffer opinions unsupported by good science. We need to reconsider the diagnostic criteria, if not the existence, of shaken baby syndrome.”

The debate over SBS is layered with unspeakable tragedy. The events happen in private — the only eyewitnesses are usually the accused and the dead. To allow a potential child abuser to go unpunished would be difficult enough to live with. But until there is sufficient medical evidence to confirm, beyond all reasonable doubt, that fatal injuries could have been caused only by foul play, to condemn an innocent, grieving parent as a murderer is infinitely worse.

Anjana Ahuja edits The Times science page. Mark Henderson returns next week