Networked Knowledge - Appeal Court Responses to Errors at Trial
The Law on Error at Trial - USA

Authors: Dr Robert N Moles and Bibi Sangha barrister

The discussion in this section is based upon the book The Riddle of Harmless Error by Roger J Traynor (1970) Ohio State University Press. Traynor was a judge of the California Supreme Court from 1940 to 1970; Chief Justice from 1964 to 1970; member of the Faculty of Law, University of California, Berkeley; Visiting Professor at the University of Virginia Law School.

It will be noted that what Judge Traynor described at the very strict English test of Stirland, has been cited with approval in the Australian case of Gipp v The Queen 1998 High Court. The point being made in Gipp and for which Stirland was cited as authority in the High Court was that an appeal may well succeed in the High Court on a point which was neither taken at trial nor in the subsequent appeal to a court of criminal appeal.

In an early English case (1807) the court considered what should be done where improper evidence has been admitted at trial. The court declared that there must be a new trial:

"… if the case without such improper evidence were not clearly made out and the improper evidence might be supposed to have had an effect on the minds of the jury."[Rex v Ball (1807) Kings Bench (emphasis added)].

Shortly after in Doe it was said that the verdict of the jury would be upheld if the appeal court thought that there was sufficient evidence to warrant the finding of the jury - independently of the evidence objected to (the “sufficiency of the evidence” test). Doe v Tyler (1830).

It was also said in Doe that the appeal court could not send the case back to a new trial - when the jury were correct on the basis of that part of the evidence which was not challenged (the “correct result” test).

A few years later (1835) the rule was brought into question again in Crease. This time it was said that the earlier views had been too broad. That view would mean that the appeal court would assume a task which was the proper function of the jury. Crease v Barrett (1835)

In another case in 1835 the Crease view was confirmed. Rutzen v Farr (1835)  

From 1835-1873, the Crease rule became one almost of automatic reversal by appeal courts when faced with trial error. The Chief Justice of the time, Lord Coleridge, said the rule was so strict that any wrong evidence would lead to a new trial. Regina v Gibson (1887)

Even minor typographical errors in the indictment were held to be sufficient to warrant a retrial.  

The Parliament, in the Judicature Act of 1873 (relating to civil trials) encouraged a return to the rule in Doe, but the appeal judges refused to take over the task of weighing the evidence. In Bray the leading judge, Lord Halsbury, declined to speculate on what the jury might have done. Bray v Ford [1896]

He said that if a reasonable jury, properly instructed might have reached another verdict, then the appeal court must reverse. This is what the lawyers call an “objective” test, and does not require evaluation of what the jury actually thought in that case.

In 1907 the Parliament in the Criminal Appeal Act tried again to achieve the goal of fewer reversals. It allowed them to dismiss the appeal if they determined that “no substantial miscarriage” had occurred, and in 1966 even the word “substantial” was removed.

Until 1970 (when Traynor’s book was published) he said that the English courts used a test that required “near certainty” that the error would not have affected the judgment:

"The appellate judges refuse to speculate as to what a jury would or should have done, and do not uphold convictions in the face of error except when no reasonable jury could have acquitted the defendant." [at p11]

He pointed out that Stirland v DPP [1944] AC 315 is the most cited case. The House of Lords set the bar very high when they said that the court must be confident that despite the error, the court would “without doubt” have convicted. Traynor said that in most subsequent decisions, the courts have not even allowed for the trial where there was an “overwhelming case” against the accused. He said they find it hard to conceive of any case where the jury would “without doubt” have convicted.

In R v Clarke [1962] dealing with an affray, six witnesses testified the defendant had not said the word “sit” to those assembled. The testimony of a seventh witness to the same effect was excluded. The court said that although seventh witness would have said the same as the other six (which the jury did not accept) they could not conclude the seventh witness would not have tipped the scale in defendant’s favour. In two cases in 1967, the House of Lords upheld the strict Stirland test.

Traynor points out that in the US, the courts initially reversed for the most trivial of errors. He said that many trials became forums for sowing reversible error in the record. He said that as a result, statutory intervention became necessary to prevent reversals where the error did not deprive a person of a fair trial. Clearly, he says, appellate review must provide a check upon arbitrary action and unfairness in trials. However, he said, we need a “harmless error” rule which avoids needless appeals on mere technicalities.

He points out that sometimes the expression used in the legislation is that the error “has not resulted in a miscarriage of justice”. However, he argues that this must be taken to mean not just the result, but the integrity of the judicial process itself. 

He says that sometimes it is argued that the test should be whether the outcome was the correct result. However, this would involve the appellate court second-guessing the jury. He says that the argument goes off course, because it assumes that a “correct” result is necessarily a “just” one. It is one thing to tolerate errors which involve mere matters of trial etiquette, or the minutiae of procedure, but it is quite another where the errors violate the rights of litigants, or debase the judicial process. Once that occurs, no one could be confident of a fair trial. The right to a fair trial includes a right to objective consideration of all proper evidence. A person is entitled not to a trial free of all possible error, but to a trial free of harmful error. He sums up by saying that the clear view in both the USA and the UK has been that where a judgment might have been influenced by error, there should be a re-trial.

He says that the evaluation of the evidence and the error by the appellate court has the virtue of saving time and money - but it does not measure up to accepted standards of fairness.

Traynor says that the arguments against such a quasi-trial on appeal, is based on the idea that it deprives the party of a trial in a trial court. The trial court can judge the temperament of a witness. The quasi appellate trial deprives the party of the opportunity to confront the witness with the issues. He says that even if we were to change the issue to “would the same result occur in a re-trial”, this would be to assume that the trial would be conducted in the same way - in the light of the new knowledge. As Traynor aptly puts it, the crucial question is not what “might happen” tomorrow on a re-run, but what “did happen” yesterday on the actual run.

He said that even if there were to be overwhelming evidence in support of a verdict, one cannot assume that to be the test of the significance of the error. If an error “left its mark” on the judgment then the verdict is tainted, and should entitle to a retrial.

He says that in the long run, there would be closer guard against error at trial if the appellate courts were alert to reverse in cases of doubt. Clearly, he says, there are no “scientific” answers as to whether an error influenced a trier of fact. It is not possible to tap the knowledge of the people (the jury) involved.

He says there is a striking difference between appellate review to determine whether an error affected the judgment (on the one hand), and whether there is substantial evidence to support a judgment (on the other).

He says that where there is a conflict in the evidence, then the appellate court can uphold the verdict if there is substantial evidence to support it.

However, he says, appellate courts not infrequently reverse judgments for error, against the weight of the evidence. In a case of error, the question is not whether there is substantial evidence to support the verdict, but whether the error affected the judgment.

He says the distinction is important. Not infrequently, a court confuses review for “error” with the “substantial evidence” test to determine if an error is harmless. The danger in this approach, he says, is that such a court tends then to consider only the evidence in support of the judgment and ignores the erroneous matter. He says that this is tied in with the courts error in applying the more probable than not test (in civil cases), instead of the “highly probable” or “almost certain” test which is applicable for criminal cases.

Obviously trial courts can distinguish between “the balance of probabilities” in civil cases and the “almost certain” standard of criminal cases. What degree of probability does the appellate court work with?

He says it is clear that unless the appellate court believes it highly probable that the error did not affect the judgment - it should reverse.

Anything less stringent would allow a judge to hold errors harmless when the results reflect their own predilections. In England, he says, the test is still “virtually certain” that the error did not affect the judgment, and this is still the test of the Supreme Court in constitutional cases. It was the one supported by Jerome Frank. In Fahy v Connecticut USA Supreme Court 1963 the Supreme Court asked if there was a “reasonable possibility” that the error influenced the result.

Traynor says that the “harmless error” rule must be seen as a substantive and not merely a procedural rule. He argues that if the test were just “highly probable”, rather than “virtually certain”, it would add to forthrightness. The more rational the test, the more likely it will be conscientiously applied. High probability is rigorous enough whilst avoiding excessive strictness. It will more likely engender a reasoned outcome.

He says that too readily affirming or reversing is just lazy review which lowers the standard of justice. In his experience, the failure to properly inquire into the matter is particularly likely in criminal cases. Whilst Judges may feel that they do not want to run the risk of a guilty defendant going free, if they do not face the risk, they face the more serious prospect of undermining the integrity of appellate review.

He says that sharper appellate review would lead to significantly less error in the trial process. Reversal for error regardless of whether it affected the judgment encourages abuse of process.

In Traynor’s view, the “highly probable” test would avoid either of the extremes.

The judgment would only be affirmed if it is highly probable that the error did not affect the outcome. It would require an examination of possible causal links between the error and the judgment.

In Fahy in 1963, the first case to suggest that a constitutional error might be harmless, the judges pointed out that there had been strong resistance to having to weigh the significance of error, because, “… they were hardly qualified to judge”.

Clearly, there are some errors which require automatic reversal without examination of the record. Examples are - a court which lacked jurisdiction over the person or the subject matter - lack of notice of charges - denial of right to a jury trial - being denied effective assistance of counsel – being denied the right to offer evidence.

Most errors require an evaluation of the effect on the judgment. Traynor says that if the test is one of being highly probable that the error did not affect the judgment, then reversal will apply, without weighing evidence, if there is a high risk of it affecting the judgment.

Examples would be prejudicial pre-trial publicity - the use of perjured testimony - suppression of evidence favourable to an accused - use of unlawfully obtained confessions. The issue, he says, is the risk of prejudice.

Traynor then made a point which in particularly applicable in the context of the Keogh case. He pointed out that in Bram v United States (1897) it was held, the prosecution cannot on the one hand offer evidence to prove guilt, and then to avoid the consequences of error, claim that the matter was not prejudicial because it did not tend to prove guilt. 

It quoted the Supreme Court of North Carolina. It is not for the State to produce evidence to secure a conviction, and then to say afterwards that the evidence did not prejudice the case against the accused. Why introduce it at all, unless it was to lay the foundation for the prosecution? The use which was made of the evidence precluded the State from saying that it was not used to his prejudice. Mr Rofe, you will remember told the Court of Appeal, and the Attorney-General told the Parliament, that it was not Dr Manock’s evidence which convicted Keogh. As we have pointed out, they cannot know that and we cannot know that. What we can say, is that it would be quite inappropriate for the DPP to state that any of the evidence which is put before the court did not influence the jury, because that is one of the tests of the evidence being admissible.

An exception might be where the evidence is merely cumulative. Where six statements or confessions were admitted and later one is found to be inadmissible. If the other five are equally as persuasive, then it may be found that the additional one was not effective on the outcome.

If an error is inherently prejudicial, then it may not be necessary to evaluate its effect.

There are some errors which are highly prejudicial “to the judicial process”. Sometimes they will be so subversive of the judicial process as to warrant reversal, irrespective of the effect in the individual case. Examples would be - error in jury selection - bias on the part of the trial judge or jury member - denial of the right to cross-examination a witness - improper exclusion of evidence - erroneously admitted evidence where not cumulative.

We would also say that the evidence of a person, not genuinely an expert, would fall into the same category. So too would evidence of a so-called expert which fails to delineate the fact that the view is not supported by other experts and has not been published. So too would be the failure to report the results of microscopical examinations to the prosecutor or to the defence.

Top of Page