SUPREME COURT OF CANADA
Citation: R. v. O’Brien, 2011 SCC 29,  2 S.C.R. 485
Her Majesty The Queen
Marty David O’Brien
Coram: McLachlin C.J. and Binnie, LeBel, Deschamps, Abella, Rothstein and Cromwell JJ.
Reasons for Judgment:
(paras. 1 to 20)
(paras. 21 to 49)
Abella J. (McLachlin C.J. and Deschamps, Rothstein and Cromwell JJ. concurring)
Binnie J. (LeBel J. concurring)
R. v. O’Brien, 2011 SCC 29,  2 S.C.R. 485
Her Majesty The Queen Appellant
Marty David O’Brien Respondent
Indexed as: R. v. O’Brien
2011 SCC 29
File No.: 33817.
2011: February 23; 2011: June 9.
Present: McLachlin C.J. and Binnie, LeBel, Deschamps, Abella, Rothstein and Cromwell JJ.
on appeal from the court of appeal for nova scotia
Criminal law — Appeals — Powers of court of appeal — Error of law — Curative proviso — Whether trial judge’s error in not excluding bad character evidence and in not specifically addressing that evidence in his reasons caused substantial wrong or miscarriage of justice — Criminal Code, R.S.C. 1985, c. C‑46, s. 686(1)(b)(iii).
A variety store was robbed by someone wearing a blue Halloween mask. The next morning, the police found such a mask, a large knife and the plastic cover from the store’s cash register nearby. A test of the DNA found on the mask matched only O, who was charged with three offences, including robbery. At a trial without a jury, the investigating officer made several brief references to O’s prior criminal record and conduct in the course of his testimony. The defence did not object to this evidence, did not request that its use be limited, and in fact used it in cross‑examination. The defence also referred to O’s reputation as a “known offender” in its closing submission. O was convicted. In his reasons, the trial judge made no reference of any kind to the character evidence. On appeal, the Crown acknowledged that it was an error for the character evidence to be heard, but submitted that since it was a harmless one, the curative proviso in s. 686(1)(b)(iii) of the Criminal Code should be applied. A majority of the judges in the Court of Appeal allowed O’s appeal and ordered a new trial, concluding that although the trial judge made no reference to the character evidence in his reasons, there was no indication that he had disabused himself of it or was otherwise aware that it should not be considered. The Crown had therefore not discharged its onus of demonstrating that the judge was not influenced by the evidence and the curative proviso was not available. The dissenting judge found the error to be a harmless one and would have applied the curative proviso.
Held (Binnie and LeBel JJ. dissenting): The appeal should be allowed and the convictions restored.
Per McLachlin C.J. and Deschamps, Abella, Rothstein and Cromwell JJ.: The hearing of the character evidence by the trial judge was a harmless error of law that caused no substantial wrong or miscarriage of justice. The trial judge said in his reasons that he relied “entirely” on the DNA evidence. That meant that he did not rely on the character evidence. Imputing such reliance into reasons that state the contrary creates a new, unchartable universe of appellate review where even if the reasons reveal a proper grasp of the facts and the law, the trial judge may nonetheless find the integrity of his or her decision undermined by the possibility that judicial silence on an issue will be interpreted as “unconscious” judicial error. A trial judge is not required to itemize every conceivable issue, argument or thought process in his or her reasons. Trial judges are entitled to have their reasons reviewed based on what they say, not on the speculative imagination of reviewing courts. Here, the trial judge expressly stated that he relied only on the DNA evidence and made no mention of the character evidence in his reasons. He was entitled to be taken at his word. The trial judge’s chain of fact, law and logic in this case was impeccable. There is nothing about his approach that suggests a subconscious subversion of his articulated thoughts.
Per Binnie and LeBel JJ. (dissenting): The curative proviso has no application. The trial judge did not intervene or otherwise express any concern about (or even awareness of) the impropriety of evidence of bad character led by the prosecution. The majority in the Nova Scotia Court of Appeal concluded that “the inadmissible evidence was no mere inadvertent slip. It went on for some pages in the transcript. It included evidence that the accused was known to the police to be involved in the very type of offence before the court.”
The Crown conceded that the DNA evidence did not, on its own, establish O’s culpability. An inferential leap had still to be made from a finding that the accused once came into contact with the mask to a finding that he actually donned the mask on the date in question to commit the robbery.
The burden on the Crown seeking to invoke the curative proviso is a substantially higher one than the requirement that the Crown prove its case beyond a reasonable doubt at trial. Important deficiencies in trial reasons cannot be excused on the basis that trial judges are presumed to know the law. Here, the trial judge’s use of the word “entirely” in relation to the DNA evidence just reflects the fact that there was no other identification evidence before him. In these circumstances, the admission of the character evidence was not harmless, since it could well have predisposed the trial judge to uncritical acceptance of the DNA evidence. If the judge had intended by the word “entirely” to distance himself from the inadmissible evidence, he would have said so. This is not a case where, despite any prejudice caused by the trial judge’s error of law, the evidence against O was so overwhelming that a guilty verdict would necessarily be the outcome of a new trial.
By Abella J.
Applied: R. v. Leaney,  2 S.C.R. 393; referred to: R. v. Khan, 2001 SCC 86,  3 S.C.R. 823; R. v. Sheppard, 2002 SCC 26,  1 S.C.R. 869.
By Binnie J. (dissenting)
R. v. Suzack (2000), 141 C.C.C. (3d) 449; R. v. Handy, 2002 SCC 56,  2 S.C.R. 908; Colpitts v. The Queen,  S.C.R. 739; R. v. Jolivet, 2000 SCC 29,  1 S.C.R. 751; R. v. Khan, 2001 SCC 86,  3 S.C.R. 823; R. v. Van, 2009 SCC 22,  1 S.C.R. 716; R. v. Mars (2006), 205 C.C.C. (3d) 376; R. v. D.D.T., 2009 ONCA 918, 257 O.A.C. 258; R. v. Samuels, 2009 ONCA 719 (CanLII); R. v. Sheppard, 2002 SCC 26,  1 S.C.R. 869; R. v. Leaney,  2 S.C.R. 393; R. v. Rahm, 2006 ABCA 111, 384 A.R. 341; R. v. Trochym, 2007 SCC 6,  1 S.C.R. 239.
Statutes and Regulations Cited
APPEAL from a judgment of the Nova Scotia Court of Appeal (Fichaud, Beveridge and Farrar JJ.A.), 2010 NSCA 61, 293 N.S.R. (2d) 78, 928 A.P.R. 78, 258 C.C.C. (3d) 358,  N.S.J. No. 385 (QL), 2010 CarswellNS 425, setting aside the convictions entered by Murphy J., 2009 NSSC 194, 279 N.S.R. (2d) 157, 887 A.P.R. 157,  N.S.J. No. 290 (QL), 2009 CarswellNS 357, and ordering a new trial. Appeal allowed, Binnie and LeBel JJ. dissenting.
Kenneth W. F. Fiske, Q.C., and William D. Delaney, for the appellant.
Robert Gregan and Pavel Boubnov, for the respondent.
The judgment of McLachlin C.J. and Deschamps, Abella, Rothstein and Cromwell JJ. was delivered by
 Abella J. — A man wearing a blue Halloween mask and holding a large, butcher-style knife came into a variety store and demanded money from the woman working in the store that night. When she opened the cash register, the robber scooped money from the till and ran out of the store. His actions were caught by the store’s security camera.
 The next morning, police found a number of items in the neighbourhood: a blue Halloween mask, a large knife, and what turned out to be a plastic cover from the store’s cash register. A test of the DNA found on the blue mask matched only one person: Marty David O’Brien. This match was characterized by the DNA expert in her report as follows:
The estimated probability of selecting an unrelated individual at random from the Canadian Caucasian population . . . is 1 in 33 billion.
 Mr. O’Brien was charged and convicted of three offences, including robbery.
 At a trial without a jury before Justice John D. Murphy, nine witnesses testified. The investigating officer made brief references to Mr. O’Brien’s prior criminal record and conduct in the course of his testimony. Not only was there no objection to this evidence by the defence nor any request that it be limited to a narrative purpose; the defence used the evidence in its cross-examination of the investigating officer:
Okay. Now in your testimony you briefly stated that you knew Mr. O’Brien to be known as an offender and that’s why you sort of steered in his direction in terms of this particular incident.
 Only defence counsel, in closing submissions, referred again to Mr. O’Brien’s reputation as a “known offender”. The Crown made no mention of the character evidence in its closing submission and the trial judge made no reference of any kind to it.
 Mr. O’Brien did not testify and the defence called no evidence.
 The basis of Mr. O’Brien’s appeal to the Nova Scotia Court of Appeal was that the verdict was unreasonable. Before the hearing, the clerk of the court, at the panel’s request, wrote to counsel and asked that they be prepared to address the issue of character evidence. Mr. O’Brien had not raised this issue as a ground of appeal either in his notice of appeal or his factum.
 At the appeal, the Crown acknowledged that it was an error for the trial judge not to exclude the character evidence or qualify its purpose, but submitted that since it was not a serious error, the curative proviso in s. 686(1)(b)(iii) of the Criminal Code, R.S.C. 1985, c. C-46, should be applied.
 All three judges concluded that the verdict was not unreasonable. The majority held, however, that the trial judge made a serious error in allowing the character evidence to be heard. In its view, even though the trial judge made no reference to the character evidence in his reasons, there was no indication that he “disabused himself from that evidence or was otherwise aware that it could not be considered” (2010 NSCA 61, 293 N.S.R. (2d) 78, at para. 94). The Crown had, as a result, not discharged its onus of demonstrating that the judge was not influenced, “consciously or unconsciously”, by the impugned prejudicial evidence (para. 95). The curative proviso was therefore not available. The appeal was allowed and a new trial ordered.
 The dissenting judge, Fichaud J.A., would have dismissed the appeal. In his view, the error in hearing the investigating officer’s propensity evidence was not a serious one and caused no substantial wrong or miscarriage of justice. He concluded:
The trial judge’s written reasons satisfy me affirmatively that the improper evidence had no impact. This, in my view, satisfies the Crown’s burden under the proviso. The judge’s words that he relied “entirely on the DNA evidence” to connect Mr. O’Brien to the robbery exclude any imputation to the judge of a veiled line of reasoning sourced in [the investigating officer’s] problematic testimony. My colleague does not explain how such a veiled line of reasoning can co-exist with the judge’s clear statement that he relied “entirely on the DNA evidence.” My colleague says that if the judge had “arrived at his conclusion by expressly relying on evidence untainted by the impugned evidence”, he might take a different view of the proviso. By my reading of the decision, that is what the judge did. . . .
. . . Nothing in the decision suggests, even obliquely, that [the investigating officer’s] improper character testimony figured in the identification. [paras. 151-52]
The curative proviso was therefore available. I share this view.
 The issue before us was whether the trial judge’s error in hearing “bad character” evidence — and his related failure to specifically address that evidence in his reasons — was a harmless one, or whether it was outside the borders of the curative proviso.
 At trial, the only issue was identification. The trial judge cautioned himself about the need to approach such evidence with care, then concluded (2009 NSSC 194, 279 N.S.R. (2d) 157, at paras. 6-7):
We then have the circumstantial evidence of a blue mask being found the following morning a short distance from the store, on a route which leads toward the car. We also have the knife and the cash register part found the following day close to the blue mask. I am satisfied that the circumstantial evidence is strong and sufficient to prove beyond a reasonable doubt that there is a connection between the blue mask and the robbery of the convenience store . . . on the previous evening.
Given the fact that the video clearly shows someone wearing a blue mask, or the blue and black mask, it clearly shows the use of a knife, and the evidence from the person who was in charge of the cash at the store, who returned to the store that night, clearly links the cash register part found to the cash register, because her notations and notations of others were on it. I am satisfied that the evidence establishes that the blue mask was associated with the robbery, and I am convinced of that beyond a reasonable doubt.
The investigating officer’s character evidence played no role in this finding.
 Turning then to the DNA evidence, the trial judge observed:
The question then becomes whether the mask is connected to the accused, to Mr. O’Brien beyond a reasonable doubt, and that depends, as I have indicated, entirely on the DNA evidence. [Emphasis added; para. 8.]
 After reviewing the DNA evidence for several paragraphs, the trial judge concluded that it “clearly ties” Mr. O’Brien to the mask, based on the irrefutable match between his DNA and the DNA found in the mask (para. 13). All this led the trial judge to conclude:
. . . the DNA evidence establishes the accused’s identity as the robber beyond a reasonable doubt, ties him to the blue mask which was found, which I have indicated, based on circumstantial evidence, is clearly the mask that was used in the robbery. [para. 16]
 The propensity evidence from the investigating officer clearly played no part in the convictions. As Fichaud J.A. noted, not only did the trial judge expressly state that he relied entirely on the DNA evidence to identify Mr. O’Brien, his detailed reasons confirm this exclusive reliance. This evokes the words of McLachlin J. in R. v. Leaney,  2 S.C.R. 393, at pp. 415-16, where she noted that when a trial judge expressly arrives at a conclusion on the critical issue “independently of the inadmissible evidence . . . no unfairness can be said to arise, nor has there been a miscarriage of justice” (see also R. v. Khan, 2001 SCC 86,  3 S.C.R. 823, at para. 30).
 The trial judge said in his reasons that he relied “entirely” on the DNA evidence (para. 8). That meant that he did not rely on the character evidence. Imputing such reliance into reasons that state the contrary creates a new, unchartable universe of appellate review where even if the reasons reveal a proper grasp of the facts and the law, the trial judge may nonetheless find the integrity of his or her decision undermined by the possibility that judicial silence on an issue will be interpreted as “unconscious” judicial error.
 A trial judge has an obligation to demonstrate through his or her reasons how the result was arrived at. This does not create a requirement to itemize every conceivable issue, argument or thought process. Trial judges are entitled to have their reasons reviewed based on what they say, not on the speculative imagination of reviewing courts. As Binnie J. noted in R. v. Sheppard, 2002 SCC 26,  1 S.C.R. 869, at para. 55, trial judges should not be held to some “abstract standard of perfection”.
 The trial judge was entitled to be taken at his word. His chain of fact, law and logic in this case was impeccable. I see nothing about his approach that suggests a subconscious subversion of his articulated thoughts.
 I therefore agree with Fichaud J.A. that the admission of the improper character evidence was a harmless error of law that caused no substantial wrong or miscarriage of justice.
 I would allow the appeal and restore the convictions.
The reasons of Binnie and LeBel JJ. were delivered by
Binnie J. (dissenting) —
 In this case, the Crown invokes the magic of DNA evidence to salvage convictions obtained in what all the judges of the Nova Scotia Court of Appeal agreed was an unfair trial.
 The respondent was convicted as the “masked robber” of a convenience store in Amherst, Nova Scotia. The only live issue was identification. There was no evidence, circumstantial or otherwise, linking him to the crime except for traces of DNA on one of two masks later found abandoned near the scene of the crime. The unfairness arose because the police, in the absence of such other evidence, decided to tell the court all about the respondent’s reputation as a habitual criminal in relation to other matters unconnected to the offences for which he was charged.
 The trial judge let in the propensity evidence without comment and gave no hint in his reasons that he recognized that the police branding of the respondent as a career criminal was against basic principles of trial fairness. Trial judges may be presumed to know the law but occasionally they misapply it. Otherwise, there would be no need for appellate courts.
 The respondent, now sentenced to six years and six months in prison, is entitled to know that the trial judge recognized the danger of propensity evidence and consciously put it aside. The trial judge’s reasons provide no assurance on that point (2009 NSSC 194, 279 N.S.R. (2d) 157).
 A majority of my colleagues conclude that the trial judge’s use of the word “entirely” in relation to his reliance on the DNA evidence is enough to render the unfairness of the trial harmless. To do so, in my view, ignores the context of the trial as a whole. The reliability of DNA science is well established, but problems in the handling of DNA samples and management of the test results have given rise to a number of wrongful convictions where (as here) there was not a shred of other probative evidence to connect an accused to the crime. DNA evidence is not a sufficient substitute for a fair trial in the circumstances disclosed in this record. The curative proviso in s. 686(1)(b)(iii) of the Criminal Code, R.S.C. 1985, c. C-46, requires the Crown, not the defence, to show that despite the unfair trial, there was no substantial wrong or miscarriage of justice. For the reasons that follow, I do not believe that the Crown has discharged its onus. I agree with the caution exercised by the majority in the court below. I would dismiss the appeal.
A. Evidence of Bad Character
 The Crown’s evidence at trial sought to portray the respondent as a “bad person” of the sort likely to commit the robbery of a convenience store. Beveridge J.A. stated that “the inadmissible evidence was no mere inadvertent slip. It went on for some pages in the transcript. It included evidence that the [accused] was known to the police to be involved in the very type of offence before the court” (2010 NSCA 61, 293 N.S.R. (2d) 78, at para. 68). He referred, at para. 70, to Doherty J.A. in R. v. Suzack (2000), 141 C.C.C. (3d) 449 (Ont. C.A.):
It has been established for over 100 years that the Crown cannot make its case by showing that the accused engaged in misconduct other than that alleged against him for the purpose of showing that the accused was the type of person who would commit the crime alleged. [para. 116]
Here, however, the police investigator, Sgt. Blakeney, testified as follows:
So just at this particular time in 2004 in October there was . . . Marty O’Brien was thought to be involved in a number of incidences [sic] within Amherst and within the county . . . . [Emphasis added; A.R., vol. II, at p. 183.]
The officer referred to the “common knowledge” of the respondent’s criminal character:
I’ve known Marty O’Brien and he’s known me for years. I don’t know, maybe 10 or maybe 15, 20 years, and I know his . . . I don’t know everything he does, but I know his criminal background, and it was common knowledge within our office and within the RCMP detachment in Amherst that Marty O’Brien was an active criminal who was traveling Amherst . . . Cumberland County area, and we were aware that Marty was involved in these activities. [A.R., vol. III, at p. 3]
This general “evidence” of criminality was unrelated to the offences as charged. The policeman continued:
So I went to the county jail, and I spoke to Marty O’Brien with regards to his being involved and he denied any involvement in it whatsoever. So I . . . after this, I checked . . . later on it became known that Marty O’Brien was convicted in court for a number of offences and a DNA order was issued to be taken on Marty O’Brien. So I continued to monitor DNA orders. [Emphasis added; A.R., vol. II, at p. 184.]
 It is true that counsel for the defence did not object. Nor did the trial judge intervene or otherwise express any concern about (or even awareness of) the impropriety of the evidence. Nevertheless, both the majority and the minority in the court below found the evidence of bad character to be wholly inadmissible and the trial on that account to be unfair. As this Court observed in R. v. Handy, 2002 SCC 56,  2 S.C.R. 908, at para. 139:
It is frequently mentioned that “prejudice” in this context is not the risk of conviction. It is, more properly, the risk of an unfocussed trial and a wrongful conviction. The forbidden chain of reasoning is to infer guilt from general disposition or propensity. The evidence, if believed, shows that an accused has discreditable tendencies. In the end, the verdict may be based on prejudice rather than proof, thereby undermining the presumption of innocence enshrined in ss. 7 and 11(d) of the Canadian Charter of Rights and Freedoms.
Propensity evidence is designed to predispose the trier of fact to accept more readily as credible the guilt of the accused than would otherwise be the case. As mentioned, Beveridge J.A. noted that its use in this case “was no mere inadvertent slip” (para. 68).
B. Evidence of Identification
 The identification of the respondent as the perpetrator rested on two strands of evidence.
(1) The Red Car
 At about the same time as the robbery, a red car was seen by a witness leaving the general area where the mask and other items were subsequently found. The respondent’s former girlfriend testified that she owned a red car to which, during the relationship, the respondent had “access”. She did not say the relationship was continuing as of the date of the robbery or that he had continuing access. The witness was shown pictures of the former girlfriend’s car, but did not identify this car as the one she saw, or provide any identifying features about it except that the car she saw was red. She could not even say whether the car in the photo was the same or close to the same colour of red. The trial judge correctly concluded that the “red car” evidence was essentially unhelpful.
(2) The DNA Evidence
 A forensic expert recovered DNA from the two abandoned face masks and testified that in the case of one of the masks, the DNA matched that of the respondent. Evidence of a “match” does not, of course, prove guilt of the offence. The Crown acknowledged that the DNA evidence in this case was the equivalent to that of a fingerprint in the sense that it established that at some point in time the respondent had handled or even worn the mask. It did not, without more, establish that the respondent was the masked robber on October 11, 2004.
 The trial judge said that he relied “entirely” on the DNA evidence to link the respondent to the mask, which the trial judge concluded (on other evidence) was the mask used in the robbery. However, an inferential leap must still be made from a finding that the respondent once came into contact with the mask to a finding that he actually donned the mask to commit the robbery. DNA evidence alone does little to bridge this analytical gap — unless one was also given the information from Sgt. Blakeney that it is not just any random member of the community whose DNA has been found on the mask, but the DNA of a career criminal who was known to be implicated in a string of similar incidents in the same area around the same time.
C. The Curative Proviso
 Section 686(1)(b)(iii) permits an appellate court to uphold a conviction despite error of law committed by the trial judge where that error has not led to a substantial wrong or miscarriage of justice. The error of law in this case was the admission of the evidence portraying the respondent as a career criminal — exactly the sort of person one might expect robbed this convenience store.
 The “curative proviso” is to be applied only in those cases where the outcome of the trial, irrespective of the error, would necessarily have been the same (Colpitts v. The Queen,  S.C.R. 739). The onus of establishing that this high standard is met is on the Crown: R. v. Jolivet, 2000 SCC 29,  1 S.C.R. 751; R. v. Khan, 2001 SCC 86,  3 S.C.R. 823; R. v. Van, 2009 SCC 22,  1 S.C.R. 716.
 The curative proviso may be applied, according to its modern interpretation, where the Crown can establish on a balance of probabilities that the legal error is harmless, in the sense that no prejudice has resulted, or that, despite any prejudice caused by the trial judge’s error of law, the evidence against the accused is so overwhelming that a trier of fact would inevitably convict: Khan, at paras. 26-31; Jolivet, at paras. 48-54; Van, at paras. 34-36.
 In the application of the curative proviso on the facts of this case, I do not think that it is sufficient to rely on the trial judge’s silence on the issues of the inadmissible police evidence. Nor should the Court’s refusal of a new trial hang on the thread of the trial judge’s use of the word “entirely” which — it seems to me — just reflects the fact that there was no other identification evidence before him. If he had intended by the word “entirely” to distance himself from the inadmissible propensity evidence, I expect he would have said so.
 My concern does not “impute” flawed reasoning to the trial judge on the basis of “speculative imagination” (reasons of Abella J., at paras. 16-17). Rather, it reflects the general principle that the Crown, not the defence, bears the burden of showing that the improper evidence was “harmless”. In my respectful view, Abella J.’s reasoning is tantamount to reversing the Crown’s burden of proof.
 The respondent argues that the evidence of bad character was prejudicial because it predisposed the trial judge to uncritical acceptance of the DNA evidence. The submission on his behalf was that “[t]he impact of DNA evidence found on the Halloween mask could have been dangerously inadvertently exaggerated by the judicial knowledge of Mr. O’Brien’s ‘proclivities’” (R.F., at para. 28).
 The Crown’s response is that the trial judge was not affected by the evidence of bad character because he didn’t mention it in his reasons. However, as noted by Beveridge J.A., the trial judge gave no indication that he was even aware of the problem — either when the evidence was being adduced or in his reasons for judgment.
D. Was the Wrongful Admission of the Evidence of Bad Character “Harmless”?
 In this case, the majority in the Court of Appeal declined to apply the curative proviso in part because there was no other evidence linking the respondent to the crime. Reference was made to R. v. Mars (2006), 205 C.C.C. (3d) 376 (Ont. C.A.), where Doherty J.A., dealing with fingerprint evidence, wrote:
Fingerprint evidence will almost always afford cogent evidence that the person whose fingerprint is left on the object touched that object. However, the ability of the fingerprint evidence to connect an accused to the crime charged will depend on whether there is other evidence capable of establishing that the accused touched the object at the relevant time and place so as to connect the accused to the crime. [Emphasis added; para. 19.]
See also R. v. D.D.T., 2009 ONCA 918, 257 O.A.C. 258. In R. v. Samuels, 2009 ONCA 719 (CanLII), by contrast, Mars was distinguished on the basis that there was “other evidence” linking the accused to the crime (para. 17).
 In the court below, Beveridge J.A. noted: “The Crown here concedes that the fact DNA was found on the blue/black mask does not, on its own, establish the . . . culpability [of the accused]” (para. 24).
The presence of the appellant’s DNA on the mask is equivalent to a fingerprint. The Crown’s expert testified that only small amounts of DNA are sufficient, and can be present for years. In addition, the thrust of the Crown’s expert evidence was that DNA can be transferred or deposited by handling. [para. 25]
 In these circumstances, I do not agree with Abella J. that admission of the evidence blackening the character of the respondent was harmless. The failure of the trial judge’s reasons to address the problematic police evidence is a deficiency that cannot simply be brushed off by reference to his use of the word “entirely”. The Crown argues that trial judges are presumed to know the law. However, the argument that important deficiencies in trial reasons can be excused on the basis that trial judges are presumed to know the law was rejected as a general one-size-fits-all excuse in R. v. Sheppard, 2002 SCC 26,  1 S.C.R. 869, and should be rejected here as well.
E. The Crown’s Reliance on the Leaney Case
 The Crown invokes R. v. Leaney,  2 S.C.R. 393. In that case, inadmissible opinion evidence was received from police officers but the curative proviso was nevertheless applied, per McLachlin J., at p. 415:
Given the trial judge’s clear statement that he arrived at his conclusion as to identify independently of the evidence of the police officers, their evidence assumes the character of mere surplusage, which does not vitiate the judge’s conclusion that Leaney was one of persons shown on the video screen. [Emphasis added.]
In my view, there is a significant difference between appellate tolerance of inadmissible opinion evidence in circumstances where the trial judge found that the opinion evidence was unnecessary to enable him to appreciate what he could see with his own eyes on a video (and the inadmissible opinions thus characterized as “mere surplusage”) and prejudicial evidence of bad character. The content of the officers’ testimony in Leaney was itself unobjectionable. The problem was that it was not their place to utter it. Here, in contrast, it is the information itself, and not just its source, that is objectionable.
 In my view, the Crown should not have the benefit of the curative proviso because the trial judge’s reasons demonstrated no awareness of the problem of the police testimony. The situation, and thus the outcome, was different in R. v. Rahm, 2006 ABCA 111, 384 A.R. 341, where Conrad J.A. held:
We are troubled by the prosecutor’s persistent line of questions relating to bad character but, in the end, we are satisfied that no miscarriage of justice resulted from the introduction of this character evidence. This was not a jury case. During the course of the trial, the trial judge made it clear that he understood that bad character was not admissible where character had not been placed in issue. [Emphasis added; para. 13.]
No such understanding is evident in the trial reasons in this case.
 In these circumstances, I do not believe that the trial judge’s error of law can be dismissed as “harmless”, and, unless we can make an affirmative finding to that effect, the respondent is entitled to a new trial.
F. The Evidence of Guilt Was Not Overwhelming
 The curative proviso may also be applied when, despite an error of law that is not harmless, the case against an accused is so overwhelming that a guilty verdict would be the only reasonable outcome of a new trial.
 In the Court of Appeal, the Crown did not even argue that its case was so strong that a trier of fact would inevitably convict. Nor did the dissenting judge make any such suggestion. In my view, these concessions were correct.
 It is an exceptional thing to deny an accused the new trial to which he is otherwise entitled because of an appellate court’s conjecture that the error was “harmless” or that any reasonable trier of fact, properly instructed, would simply find the evidence of guilt to be overwhelming.
 On this point, the majority of our Court stated in R. v. Trochym, 2007 SCC 6,  1 S.C.R. 239, per Deschamps J., at para. 82, that the burden on the Crown seeking to invoke the curative proviso “is a substantially higher one than the requirement that the Crown prove its case ‘beyond a reasonable doubt’ at trial”.
The standard applied by an appellate court, namely that the evidence against an accused is so overwhelming that conviction is inevitable or would invariably result, is a substantially higher one than the requirement that the Crown prove its case “beyond a reasonable doubt” at trial. This higher standard reflects the fact that it is difficult for an appellate court . . . to consider retroactively the effect that, for example, excluding certain evidence could reasonably have had on the outcome. [Emphasis added; para. 82.]
In my view, the Crown cannot establish that its case against the respondent was “overwhelming”.
 I agree with the majority opinion in the court below that the curative proviso has no application, and accordingly I would dismiss the appeal.
Appeal allowed, Binnie and LeBel JJ. dissenting.
Solicitor for the appellant: Public Prosecution Service of Nova Scotia, Halifax.
Solicitor for the respondent: Nova Scotia Legal Aid, Amherst.