FINLEY, J.

Carl Knecht is suing as guardian ad litem for his son Timothy, age 3, to recover damages for personal injuries sustained when the minor ran across a residential street and was struck by a car driven by the defendant, Pearl Marzano. The trial jury returned a verdict for the defendant. Thereafter, the trial judge heard argument of counsel and, acting under Rule of Pleading, Practice and Procedure 59.04W(9), entered an order granting a new trial as follows:

"ORDERED, ADJUDGED AND DECREED that the plaintiff be, and he hereby is, awarded a new trial. The Court's reasons for granting the new trial are as follows:

"That substantial justice was not done. The Court is unable to point to any precise or specific matter of law or fact on which to base its ruling, but the Court has the strong feeling that substantial justice was not done, based upon its whole impression of the trial." (Italics ours.)

On appeal the issue is whether the above-quoted order complies with Rule 59.04W, and thus whether granting the new trial was error.

[1] Manifestly, the trial judge's order does not begin to comply with the requirements of Rule 59.04W, that

". . .

"In all cases wherein the trial court grants a motion for a new trial, it shall, in the order granting the motion, give definite reasons of law and facts for so doing. . . ."

Reversal of the order could be based solely upon noncompliance with the long standing, well known rule. Johnson v. Department of Labor Industries, 46 Wn.2d 463, 466, 281 P.2d 994 (1955). However, counsel for respondent argues against this strongly. He urges that in our appellate review of the matter consideration be given to the reasons for granting the new trial as stated by the trial judge in his oral decision. We look somewhat hesitantly upon the statements in the oral decision of the trial judge, because he refused to put them in his formal order; but we will review and consider the statements, because of the importance of the procedural problem and the doubts raised regarding a fair trial to this permanently and severely injured child.

[2] In his oral discussion the trial court assigned two reasons in support of his feelings respecting a failure of substantial justice in the instant case. First of all, he indicated that the jury ignored the fact that the defendant left 70 feet of skid marks before she hit the plaintiff. Apropos of expert testimony and evidence of actual experiments tending to show that a car traveling the 25 mile-per-hour speed limit would leave only 32 to 18 feet of skid marks with the wheels locked in a panic stop, the trial judge was convinced that the evidence was overwhelming of excessive speed by the defendant. Second, the trial judge alluded to a feeling of considerable doubt on his part respecting the reliability of the defendant's testimony.

Unfortunately, the above reasons or opinions expressed by the trial judge provide little or no assistance respecting appellate review of this case. He simply disagreed with the jury. If we had been in the court room or the jury box, we might have agreed with the trial judge's first statement of opinion or judgment that the defendant was speeding. But, under our decisions since the advent of Rule 59.04W, this is not the test applicable upon appellate review respecting the juristic problem here involved. Absent more tangible, convincing and adequate reasons supporting the trial judge's first theme respecting the failure of substantial justice, we shall now determine whether there was sufficient evidence to uphold the verdict. Greenwood v. Bogue, 53 Wn.2d 795, 337 P.2d 708 (1959); Davenport v. Taylor, 50 Wn.2d 370, 311 P.2d 990 (1957). The defendant strongly contested the experimental evidence offered by the plaintiff and introduced evidence which showed that only the wheels on the right side of defendant's car locked and produced the skid marks. The defendant also argued that the 70 feet of skid marks left by the defendant should include a deduction of 20 feet for the length of the car, as it was possible for the rear wheel to lock first upon application of the brakes, and for the front wheel to be locked only toward the end of the process. The jury apparently accepted this evidence. We are not constrained to hold it was insufficient to uphold the verdict. Nor can we justifiably affirm the granting of a new trial on the trial judge's indefinite feeling that the defendant's testimony was not credible. Credibility is for the jury. The trial judge has only demonstrated personal disagreement with the jury — we must reverse.

We could probably terminate this opinion with the foregoing dispositional discussion. However, the trial courts and this court have been troubled frequently respecting the application and the function of Rule 59.04W(9), which allows the trial judge to grant new trials where substantial justice has not been done. See footnote 2, Sullivan v. Watson, 60 Wn.2d 759, 764, 375 P.2d 501 (1962); and Trautman, New Trials for Failure of Substantial Justice, 37 Wn. L. Rev. 367 (1962). Consequently, further elaboration may be appropriate. Professor Trautman, in the above-cited article, has traced the development of Rule 59.04W(9), which may be summarized briefly as follows. Before the adoption of the rule, a trial judge had inherent discretionary power to award new trials on the ground that there had been a failure of substantial justice, and this court was bound to affirm the granting of a new trial unless there was no case for a jury or there was no evidence to support a verdict other than the one rendered. The resulting limitation upon the performance of any actual appellate function was referred to as an "iron curtain" obstacle to review in Coppo v. Van Wieringen, 36 Wn.2d 120, 217 P.2d 294 (1950). That case indicated to the bar that the court was about to make rule changes affecting this matter of practice and procedure. In 1951, the granting of a new trial for failure of substantial justice was added to the previous eight statutory grounds for granting new trials, forming the basis of what is now substantially Rule 59.04W. But at that time the rule inaugurated a new requirement, referred to hereinbefore, that the trial judge give "definite reasons of law and facts" to support the granting of a new trial. While this requirement applies to all nine grounds for granting new trials under the new rule, it was particularly aimed at correcting the lack of review identified in the Coppo case in relation to the ground that substantial justice had not been done. The pendulum then swung. As Professor Trautman writes:

"The ground of failure of substantial justice has not fared well since the adoption of the rule requiring definite reasons of law and facts. Instead of a situation where the supreme court rarely reversed the grant of a new trial, as existed prior to the rule, there is now a situation where the grant of a new trial on substantial justice grounds is rarely affirmed. . . ." Trautman: New Trials for Failure of Substantial Justice, supra, p. 375.

Albeit, the pendulum recently shows a tendency of swinging back to a middle ground, indicative that in appropriate cases a grant of a new trial for failure of substantial justice will be affirmed. See: Dipangrazio v. Salamonsen, 64 Wn.2d 720, 725, 393 P.2d 936 (1964); Cyrus v. Martin, 64 Wn.2d 810, 812, 394 P.2d 369 (1964); see, also footnote 2, Sullivan v. Watson, supra.

The difficulty of formulating a realistic basis for a middle ground between the two extremes relative to review of a grant of a new trial for the failure of substantial justice can be seen in the above history of the problem. While we are aware of the forceful arguments or recommendations for a change in Rule 59.04W(9), or its interpretation and application in Professor Trautman's interesting article in the Washington Law Review, supra; nevertheless, it seems to us that the sprightly barbs aimed at Rule 59.04W(9) fall short of the mark, inadvertently perhaps, in gauging the dimensions of the target problem. Involved here is more than just appellate court versus trial court functions, juristic, or other relationships. It is not just a matter of according discretion to trial court judges and testing the propriety of the exercise of such discretion in specific instances against the appellate touchstone — " manifest abuse of discretion." Although "manifest abuse of discretion" is a formidable and inferentially an effective appellate standard or mechanism for testing the exercise of trial judge discretion in some areas, its use or function seems debatable, to say the least, in the absence of a requirement that produces objectively assessable criteria, reasons or facts respecting the granting of a new trial, rather than ones assessable only subjectively in terms of describing or revealing nothing more than the feelings or hunches of the trial judge to the effect that "substantial justice has not been done." Stated somewhat differently, discretion equated only with the feelings and hunches of the trial judge is not amenable to objective evaluation and appellate review, for the end result would be nonreviewable trial judge discretion — in essence, no appeal whatsoever.

[3] But again, thinking in terms of dual dimensions, or bilaterally, trial judge versus appellate court functions or relationships, is not a full approximation or evaluation of the problem. There is a third dimension or factor: namely, the jury. With this in mind, the problem is now triangular, or three dimensional, involving (1) trial judge, (2) jury, and (3) appellate court functions and relationships.

It seems not irrelevant that in our jurisdiction constitutional mandate elevates the factual decision-making function of the jury and limits or preempts this function from the trial judge by preventing his commenting upon and emphasizing or orienting the evidence in terms of the final jury decision. This philosophy of the constitution and the decisional law in our state is well settled in describing the province or the factual decision-making function of the jury. On occasion, we have had the conviction that a trial judge was disguising a personal disagreement (possibly subconscious) with a jury result in his conclusion that substantial justice was not done. We have thought that this was a clear invasion of the province of the jury, and we have not hesitated to guard and to attempt to delineate clearly the boundary between the factual decision-making functions of the jury and the distinguishable functions of both the trial judge and the appellate court.

However, it is desirable to avoid becoming too mechanistic in approach. We can foresee and understand occasions when a trial judge may say, "The jury verdict is supported by sufficient evidence, but X and Y extra-record factors, singly or in combination, caused the jury to give far too much consideration to that evidence, which resulted in an unfair trial, and consequently a new trial must be granted because substantial justice has not been done." In other words, on rare occasions, it might be possible for a trial to be derailed, resulting in the failure of substantial justice, when, on appellate review and under the present application of the rule, we would affirm the jury verdict. In any event, what we must insist on, and what Rule 59.04W requires, is an adequate explanation of the various extra-record factors which caused the jury to make a gross error. If these reasons are present in the order granting a new trial, then this court will undertake its duty to accord a fair review to the determination of the trial judge who witnessed the derailing of the normally adequate trial system. In this way we will recognize and give proper effect to the functions of the trial and appellate courts and the jury. Most importantly, if the above procedure is followed, then our fears that trial courts could attenuate or water down the jury function will be reasonably obviated by reasonably operable appellate review, and the trial court and this court will not be parties to "jury shopping"; rather, we will be giving the parties litigant their just due — a fair trial. Since a jury must still make the ultimate determination, the judge will not be invading the jury box in granting a new trial for failure of substantial justice upon proper and stated reasons.

We might say, in passing, that the granting of new trials for the lack of substantial justice should be relatively rare, especially since Rule 59.04W gives eight other broad grounds for granting new trials. We do have a distinct feeling that in the granting of new trials some use of the ground, "That substantial justice has not been done," may have approximated the cry of "Wolf! Wolf!", with the age-old reaction to that cry. Limited use of this ground under the circumstances outlined above should give this needed component of trial administration new, constructive vigor.

In conclusion, it is our opinion that the trial court failed to give any extra-record reasons which would explain the claimed gross error of the jury. Since there is sufficient evidence to uphold the verdict, we can only conclude that the trial court disagreed with the jury, and this is not enough. The order granting a new trial is reversed, and judgment for the defendant will be entered.

OTT, C.J., WEAVER and HAMILTON, JJ., and BARNETT, J. Pro Tem., concur.