What happened
Peter Rex Dansie was convicted by Lovell J sitting alone in the Supreme Court of South Australia of the murder of his wife on 16 April 2017. The couple had been married for over 40 years. The wife had suffered a stroke in 1995 that left her with long-term physical and cognitive disabilities. By 2015 she was wheelchair-dependent and resided permanently in a nursing home. On the day in question Dansie collected her from the nursing home and drove her to the South Parklands in Adelaide. He positioned her wheelchair near a pond slightly more than one metre deep. He later called emergency services. When ambulance officers and police arrived the wife's body was face down in the pond, the wheelchair was also in the pond, and Dansie was wet to his waist. There were no independent eyewitnesses.
The prosecution case was that Dansie had deliberately pushed the wheelchair into the pond intending to drown his wife. The defence case was that the wheelchair had entered the water accidentally while Dansie was attempting to manoeuvre it away from the edge. The issue was therefore whether the prosecution could prove murder so as to exclude accidental drowning as a reasonable possibility. The evidence was largely uncontested and presented in documentary form or as transcripts of unchallenged testimony from an earlier aborted trial. It included extensive material about the couple's relationship and financial circumstances, the wife's condition, the topography of the pond, Dansie's movements that day, his call to emergency services, his appearance and conduct when officers arrived, his police interviews, the results of searches of his car and home (including a computer showing internet searches about funerals the previous month), and intercepted telephone calls. Lovell J noted there was little dispute about primary facts; the case turned on inferences. His Honour drew adverse inferences about Dansie's credit from the interviews, found his account of events at the pond implausible, inferred a changed and non-caring relationship, concluded his emergency call showed lack of genuine concern, accepted evidence of preparatory acts, and found two interconnected motives (financial and relational). He concluded the only rational inference was that Dansie deliberately pushed the wheelchair with intent to kill and convicted him of murder.
Dansie appealed to the Full Court of the Supreme Court of South Australia sitting as the Court of Criminal Appeal. He advanced a ground that the verdict could not be supported having regard to the evidence, invoking s 158(1)(a) of the Criminal Procedure Act 1921 (SA). The Full Court unanimously rejected grounds alleging specific errors in Lovell J's reasons. By majority (Parker and Livesey JJ, Nicholson J dissenting) it also rejected the unreasonableness ground and dismissed the appeal. Livesey J (with whom Parker J agreed) reviewed the primary facts as found by Lovell J, agreed with the inferences drawn, and expressed the view that there was a clear pathway to proof of guilt. However, his Honour's analysis was framed as an inquiry into error in the trial judge's decision, with deference to the primary role of the trier of fact and reliance on the formulation in Plomp v The Queen. Nicholson J, in contrast, performed a detailed independent assessment of the circumstantial case and would have allowed the appeal.
By special leave the High Court heard Dansie's further appeal on the single ground that the majority had misinterpreted and misapplied the approach required by M v The Queen as applied in Filippou v The Queen. The High Court (Gageler, Keane, Gordon, Steward and Gleeson JJ jointly) unanimously allowed the appeal, set aside the Full Court's order, and remitted the matter for rehearing. The Court emphasised that the majority had not undertaken the independent assessment of the evidence required by the M test but had instead searched for error and treated certain inferential questions as beyond its proper role.
Why the court decided this way
The High Court decided the case on the basis that the majority in the Court of Criminal Appeal had misunderstood the function an appellate court must perform on an unreasonableness ground. That function is not to review the trial judge's reasons for error in the manner of an appeal by way of rehearing on factual findings. Rather, it is to ask for itself whether, upon the whole of the evidence, it thinks it was open to the trier of fact to be satisfied beyond reasonable doubt of guilt. The Court grounded this conclusion squarely in the joint judgment in M v The Queen, which it said had settled the understanding of the appellate function under common-form criminal appeal statutes. At paragraph 4 the Court reiterated that the question is one of fact answered by the appellate court's own independent assessment of the evidence, both sufficiency and quality.
The joint judgment stressed that this assessment must not disregard the jury (or judge) as the body entrusted with primary responsibility or the advantage that body had in seeing and hearing witnesses. However, the Court explained (drawing on the carefully crafted passage in M at 494-495) that a reasonable doubt experienced by the appellate court is a doubt a reasonable jury (or judge) ought to have experienced unless the trier of fact's advantage in seeing and hearing the evidence is capable of resolving it. Where the evidence on the record contains discrepancies, inadequacies or lacks probative force in a way that leaves a significant possibility an innocent person has been convicted, the appellate court must set the verdict aside. The Court noted that M had provided authoritative guidance and that earlier formulations irreconcilable with it (including Menzies J's statement in Plomp) must be treated as overtaken.
Filippou v The Queen was treated as confirming that the M test operates in exactly the same way after a judge-alone trial. The appellate court still performs an independent assessment and will set aside the conviction if its own view of the evidence leaves a reasonable doubt that the trial judge's advantage cannot resolve. At paragraph 7 the High Court warned that trial reasons in a judge-alone case must be approached with circumspection lest they divert the appellate court from that independent task. The court is entitled to treat undisputed findings as accurate but the ultimate question remains whether the appellate court's assessment of the totality of the evidence leaves it with a reasonable doubt it cannot assuage by reference to the trial judge's advantage.
In the present case the prosecution was circumstantial, the evidence was largely uncontested and in transcript or documentary form, and Dansie did not give evidence. The High Court observed at paragraph 8 that the trial judge's advantage was therefore slight. The majority's reasons, however, were directed from beginning to end to the detection of error in Lovell J's decision and to deference to his primary role as fact-finder. They relied on Plomp, equated the appellate question with Menzies J's formulation, stated it was not for the Court to determine whether the only rational inference was guilt, described inferential questions as primarily and classically for the trier of fact, and declined to "dwell upon" defence arguments about inferences consistent with innocence. The majority satisfied itself that there was a "clear pathway to proof of guilt" but did not independently weigh whether reasonable hypotheses consistent with accident remained open. Nicholson J, by contrast, had correctly recognised that the assessment required him to form a view on those inferences even if they might be called "jury questions".
Because the majority had not performed the independent assessment required by M, the High Court held that they had misapplied the test. The appeal was therefore allowed and the matter remitted so that the Court of Appeal could rehear the unreasonableness ground according to law. Neither party invited the High Court to perform the M assessment itself.
Before and after state of the law
Prior to M v The Queen the law on the unreasonable verdict ground contained a variety of formulations, some of which placed greater emphasis on deference to the jury or on whether there was sufficient evidence on which a jury could convict. The High Court noted that some of these earlier statements, including the approach taken by Menzies J in Plomp v The Queen at 247 and 252, involved a degree of deference to the inference of guilt drawn by the jury that does not accord with the M test. The joint judgment in M was intended to provide authoritative guidance and to put aside differences in expression in previous cases. It required the appellate court to make its own independent assessment rather than merely ask whether there was evidence on which a reasonable jury could have convicted.
Subsequent decisions such as MFA v The Queen, SKA v The Queen, Coughlan v The Queen and Pell v The Queen had affirmed and applied the M test without diminishing its authority. SKA made clear that the test requires independent assessment of both sufficiency and quality and that so-called "jury questions" are not off-limits. Pell confirmed that nothing in Libke v The Queen had departed from M. Filippou expressly extended the same approach to judge-alone trials, requiring the appellate court to ask whether it was open to the judge to be satisfied beyond reasonable doubt, with the same qualification concerning the judge's advantage.
After Dansie the law is clearer and more uniform. The High Court has reiterated that the M test is not engaged by searching for error in a trial judge's factual findings or inferences. Even in a judge-alone case the appellate court must itself weigh the evidence, including in circumstantial cases deciding whether the prosecution has excluded all reasonable hypotheses consistent with innocence. Courts must be wary of letting the structure of written reasons for judgment divert them from that task. The advantage of a trial judge is to be assessed case-by-case and may be slight where the evidence is largely documentary or transcript-based. The decision reinforces that the appellate function on the unreasonableness ground is not a species of error-correction but a merits assessment of whether a reasonable doubt ought to have been experienced. Any formulation that suggests the appellate court should simply ask whether the trier of fact could rationally have convicted is now to be read subject to M and, where inconsistent, treated as overtaken.
Key passages with plain-English translation
At paragraph 3 the Court stated: "The function to be performed by the Court of Criminal Appeal when determining an appeal on the unreasonable verdict ground is not to determine whether there was error in the factual findings on which the trial judge relied in ultimately finding the accused guilty of the offence. The function to be performed by the Court of Criminal Appeal is to determine for itself whether the evidence was sufficient in nature and quality to eliminate any reasonable doubt that the accused is guilty of that offence." In plain English this means the appeal court does not sit to mark the trial judge's homework for mistakes; it must look at the evidence afresh and decide whether that evidence leaves any real doubt about guilt.
Paragraph 4 repeats the central question from M: the court must ask "whether it thinks that upon the whole of the evidence it was open to the jury to be satisfied beyond reasonable doubt that the accused was guilty", and this is answered by the court's "own independent assessment of the evidence". Translation: the appeal judges must form their own opinion after reviewing everything, not merely decide whether a reasonable judge or jury somewhere could have convicted.
The lengthy passage from M quoted at paragraph 5 is central: "It is only where a jury's advantage in seeing and hearing the evidence is capable of resolving a doubt experienced by a court of criminal appeal that the court may conclude that no miscarriage of justice occurred... If the evidence, upon the record itself, contains discrepancies, displays inadequacies, is tainted or otherwise lacks probative force in such a way as to lead the court of criminal appeal to conclude that, even making full allowance for the advantages enjoyed by a jury, there is a significant possibility that an innocent person has been convicted, then the court is bound to act and to set aside a verdict based upon that evidence." Plain English: if the appeal court harbours a doubt after reading the record, that doubt should have led to acquittal unless the fact that the trial judge saw and heard witnesses (in a way that cannot be replicated on appeal) explains why the doubt is unreasonable. If the evidence looks shaky on paper, the conviction cannot stand.
At paragraph 7 the Court said: "Where the trial has been by judge alone, the reasons of the trial judge must be approached by the court of criminal appeal performing that function with circumspection lest the findings of fact made by the trial judge divert the court from undertaking the requisite independent assessment of the evidence." Translation: do not let the trial judge's neatly organised reasons become a substitute for your own review of the raw evidence; treat them carefully so they do not distract you from doing the M assessment.
Finally, paragraph 11 sets out the concrete task for this case: the appellate court had to ask whether it was "independently satisfied as a result of [its] own assessment of the whole of the evidence... that the only rational inference available on that evidence was that the appellant deliberately pushed the wheelchair into the pond with intent to drown his wife and, if not, whether the satisfaction arrived at by Lovell J could be attributed to some identified advantage which Lovell J had over [it]". In plain language: read all the material yourself, decide if accident is still a reasonable possibility, and only uphold the conviction if any doubt you have can be explained by something the trial judge saw or heard that you cannot.
What fact patterns trigger this precedent
This precedent is triggered on any appeal against conviction on the unreasonable verdict ground under provisions in the common form of s 158(1)(a) of the Criminal Procedure Act 1921 (SA) or equivalent statutes. It is especially relevant where the trial was by judge alone, because the existence of written reasons creates a temptation to treat the appeal as a search for error in those reasons rather than an independent assessment. The decision applies with particular force in circumstantial cases where the ultimate issue is whether the only rational inference is guilt or whether a hypothesis consistent with innocence remains open. It is engaged whenever the evidence on appeal is substantially the same as the record before the trial judge (transcripts, documents, exhibits) and does not turn on contested oral testimony that the trial judge alone observed.
The precedent is also engaged where the prosecution case depends on inferences from primary facts that are largely undisputed, such as motive, post-offence conduct, inconsistencies in interviews, and preparatory acts. It applies whether or not the accused gave evidence. The Court noted that in the present matter the largely uncontested, documentary nature of the evidence meant the trial judge's advantage was slight; the same will be true in any similar "paper" case. Conversely, the precedent reminds courts that where live credibility contests occurred and the trial judge had a real advantage in seeing and hearing witnesses (especially the accused), that advantage must be weighed, but only to the extent it can actually resolve a doubt the appellate court harbours after its own review. Any appeal that involves debate about "jury questions", the weighing of competing inferences, or the sufficiency of circumstantial evidence will now be measured against the strictures in Dansie.
How later courts have treated it
Although the judgment itself is the authoritative statement, it makes plain how earlier authorities are to be treated. M v The Queen is to be applied according to the joint judgment's carefully crafted propositions at 494-495; earlier decisions that cannot be reconciled with it, such as the statements of Menzies J in Plomp v The Queen, are not to be followed on the formulation of the appellate question. The Court noted that Plomp had been relied upon by the majority below and that the propositions derived from it led the majority to fail to undertake the independent assessment required. Similarly, the decision makes clear that observations in R v Baden-Clay about the serious step of setting aside a conviction must be read in the context of M and do not authorise a general deference that displaces independent assessment.
The judgment treats Filippou v The Queen as having settled that the M test operates identically for judge-alone trials. Citations of SKA v The Queen, Coughlan v The Queen and Pell v The Queen are used to reinforce that the independent-assessment obligation includes weighing circumstantial evidence, considering so-called jury questions, and not reading Libke v The Queen as having weakened M. The High Court approved Nicholson J's dissenting approach in the Full Court as correctly recognising that the assessment required a view to be formed on inferences notwithstanding that they might be characterised as jury questions. In short, the decision positions M and Filippou as the governing authorities; any appellate reasoning that frames the task as error-detection, that defers to the trier of fact on the drawing of inferences without performing its own weighing, or that treats Plomp as supplying the applicable test, is inconsistent with Dansie and must be regarded as erroneous.
The judgment does not itself cite subsequent cases (being the decision that settles the point), but its repeated insistence that the M test has not diminished with time and must be applied without the distractions seen in the Full Court below signals that later courts are to treat Dansie as confirming a strict, evidence-focused appellate function that cannot be discharged by merely agreeing with a trial judge's pathway to guilt.
Still-open questions
The judgment leaves open the precise degree of advantage a trial judge may have in cases that fall between the largely documentary present matter and a case involving substantial live contested oral evidence. At paragraph 8 the Court said the advantage "will vary from case to case depending on the form in which the evidence was adduced at the trial and depending on the nature of the issues that arose at the trial". Future courts will need to articulate when an advantage is "capable of resolving" a doubt experienced on the record. The decision does not prescribe a mechanical test for that assessment.
Another open question is the extent to which an appellate court may make additional findings or draw additional inferences beyond those of the trial judge. The majority below had expressed some additional views but only to bolster a pathway to guilt. The High Court did not disapprove of that in principle but emphasised that the core task is the independent M assessment rather than supplementation of the trial reasons. Exactly how far an appellate court may go in re-characterising the evidence while still respecting the trial judge's advantage remains to be worked out in future cases.
The judgment also leaves for future consideration the interaction between the unreasonableness ground and the obligation to give reasons in judge-alone trials. While it affirms that reasons are necessary to permit appellate review under s 158(1)(b) and (c), it warns that those reasons must not be permitted to distort the separate s 158(1)(a) task. The precise forensic approach an appellate court should take when the trial reasons are comprehensive yet the appellate court harbours a doubt the reasons do not address is not exhaustively mapped.
Finally, the Court did not itself apply the M test to the evidence in this matter, noting that neither party asked it to do so. The question of what conclusion would have been reached on a correct application (and whether the circumstantial case ultimately excluded accident) is therefore left for the Court of Appeal on remittal. That rehearing will itself test how later courts operationalise the guidance given in Dansie.