TAYLOR, J.

The only question presented by this appeal is the propriety of an order granting plaintiff’s motion for a new trial for errors of law. Plaintiff filed this action for personal injuries sustained in a collision with a vehicle 'driven by defendant. After the jury brought in a verdict for defendant, the trial court granted plaintiff’s motion for a new trial because it had admitted into evidence, for the limited purpose of impeaching plaintiff’s expert witness, the results of an experiment conducted by defendant, and also because it failed to instruct the jury on the limited purpose for which this evidence was admitted. Defendant argues that the order granting the new trial cannot be sustained on either ground specified.

Before directly discussing the new trial proceedings, some reference should be made to the evidence. The accident occurred on a dry sunny morning in November 1960 at the intersection of Ross Road and Moreno Avenue in Palo Alto. Ross runs in a north-south direction and Moreno east-west; both are asphalt paved and about 36 feet wide. The intersection was controlled by a “yield right-of-way” sign on the southeast corner. A solid concrete block fence about 3% feet high runs parallel to the east side of Ross Road and parallel to the south side of Moreno. Two trees on the eastern side of Ross, just south of the intersection, were in foliage at the time of the accident.

Defendant, accompasiied by his wife, was driving north on Ross in his 1959 Dodge station wagon. As he approached the intersection, he noticed several children playing in the northwest corner but did not see the yield sign on the southeast corner or any other traffic. After looking to the right, he decreased his speed and entered the intersection at a speed of 5-10 miles per hour. When he was between one-quarter and one-half of the way into the intersection, he saw plaintiff’s car about 35-40 feet away on his right but did not stop because he felt there was sufficient clearance for her to pass. He estimated her speed to be about 40 miles per hour. When the front of his ear was in the center of the intersection, he accelerated and thus estimated his own speed at the time of impact to be *22 15-20 miles an hour. He did not apply the brakes and lost control of the car after the impact.

Plaintiff was driving west on Moreno in her 1960 Corvair. She turned onto Moreno from. Fielding, 2 blocks east of Ross. As she crossed Marshall Street, one block (about 275 feet) east of the Ross Road intersection, she was in second gear. After crossing Marshall Street, she shifted into high gear, then took her foot off the accelerator to look down Ross and saw defendant’s ear. She estimated its speed to be approximately the same as her own, about 25 miles per hour.

The impact occurred in the northeast quadrant of the intersection, about 24 feet north of the south intersection line and 10 feet west of the east intersection line. Plaintiff’s car struck defendant’s car behind the right rear wheel and caused it to make a 180 degree turn and come to rest facing south, the direction from which it had come. After the impact, plaintiff’s car moved about 1-3 feet.

The Palo Alto police officer investigating the accident indicated that plaintiff’s car left 30 feet of right front and rear wheel skid marks and 27 feet left front and rear wheel skid marks in a straight line. Defendant’s car left 78 feet of left rear and right rear wheel side skid marks and 45 feet of left front wheel side skid marks, all at an angle in a northerly direction.

Plaintiff’s expert witness, Dr. Laitone, was asked a series of hypothetical questions about the speed of both ears, based on his calculations that took into account the size, weight and year of both cars, the length and nature of the skid marks, the coefficient of friction on dry asphalt pavement, the turning of defendant’s car and the area of contact between the two cars. He testified that on the basis of his calculations, the speed of defendant’s car at the point of impact was 29 miles per hour if the brakes had not been applied, and 40 miles per hour if the brakes had been applied; the plaintiff’s speed at the time her skid marks started was about 25 miles per hour and at the point of impact, 14 miles per hour, assuming her car had moved 3 feet after the impact. He could not calculate plaintiff’s speed before her wheels locked and she began to lay down skid marks. Dr. Laitone also pointed out that his answers to all of the hypothetical questions had an error of plus or minus 10 percent.

Defendant’s expert witness estimated that plaintiff’s speed at the time of impact was 10 miles per hour, and at least 35 miles per hour when the ear started skidding, with a possible *23 error of 10 percent. He could not determine the speed of defendant’s car at the time of impact.

To impeach the testimony of Dr. Laitone that the 30 feet of skid marks left by plaintiff’s car were the result of a speed of 25 miles per hour, defendant introduced the testimony of a private investigator, Mr. Currie, who was a passenger in a 1960 Corvair, driven by defendant’s attorney, Mr. Britton, in a series of experiments. Currie testified that on two occasions, after the speedometer showed a speed of 25 miles per hour, Britton applied the brakes as hard as he could on level dry asphalt. The car left skid marks of 23 feet 7 inches and 24 feet 6 inches. In a subsequent test of the accuracy of the speedometer, a 15 percent error was found at speeds of 30 and 60 miles per hour.

On rebuttal, Dr. Laitone testified that if this 15 percent error had been taken into account, the car used in defendant’s test was apparently going at a speed of 21% miles per hour when the speedometer showed 25. If the car had, in fact, been doing 25 miles per hour, it would have left skid marks of 34 feet 4 inches and 33 feet 3 inches, respectively.

The granting or denial of a new trial is a matter resting so largely in the discretion of a trial court that it will not be disturbed except upon a manifest and unmistakable abuse. This is especially so when the discretion is used in awarding a new trial, for such au order does not finally dispose of the matter (4 Cal.Jur.2d, p. 476, § 598). Thus, the question here presented is not whether the court’s error was prejudicial but whether the court abused its discretion in concluding, as a matter of law, that the experimental evidence confused the jury and did not substantially tend to establish the fact it was offered to prove, and that it erred in failing to give an instruction on the limited purpose of the evidence in question.

Testimony relating the results of experiments concerning a disputed material fact should be admitted only if the conditions under which the experiments were made are substantially identical to those out of which the disputed fact arose (Beresford v. Pacific Gas & Elec. Co., 45 Cal.2d 738, 748 [290 P.2d 498, 54 A.L.R.2d 910]), should be of that character which will clarify rather than confuse the issue at hand (Martin v. Angel City Baseball Assn., 3 Cal.App.2d 586, 590 [40 P.2d 287]), and should not be admitted if it predominantly involves a consideration of collateral issues (Braly v. Midvalley Chemical Co., 192 Cal.App.2d 369, 380 [13 Cal.Rptr. *24 366]; Garcia v. Hoffman, 212 Cal.App.2d 530, 535 [28 Cal.Rptr. 98]). The determination of these factors is a matter largely within the discretion of the trial court, whose ruling will not be interfered with on appeal in the absence of a clear showing of abuse of that discretion (Buratti v. Phetteplace, 196 Cal.App.2d 303, 309 [16 Cal.Rptr. 500]; Grupe v. Glick, 26 Cal.2d 680, 685 [160 P.2d 832]).

In the case at bar, the conditions at the time of the accident and those at the time of the subject experiment presented many variables. Conditions are not substantially similar if they vary to such a degree that the result of the experiment will be altered. Here, the error in the speedometer giving rise to a probable speed of 21*4 instead of 25 miles per hour, could reduce the length of the skid marks by almost 10 feet. Currie could only testify that Britton applied the brakes as hard as he could. Defendant offered no evidence of the application of maximum braking pressure. Dr. Laitone indicated that the proper application of maximum braking pressure to simulate the panic-braking of a driver involved in an accident was very difficult. The use of hard rather than maximum braking pressure could distort the stopping distance by as much as 7 feet. Thus, the trial court could reasonably conclude that the variation in the speed and braking pressure destroyed the value of the experiment and made it harmful rather than helpful (People v. Carter, 48 Cal.2d 737, 759-760 [312 P.2d 665]).

Furthermore, the court was justified in concluding that the evidence of the experiment did not sufficiently establish the facts for which it was proffered. The record indicates that Dr. Laitone, plaintiff’s expert, testified, in response to a number of hypothetical questions, that on the basis of his calculations based on the skid marks at the scene of the accident, the size, weight and impact of the two cars and the coefficient of friction, a car of the same make and model as plaintiff’s would have to be traveling about 25 miles an hour to lay ’down 30 feet of skid marks. Dr. Laitone’s testimony concerned the speed of an automobile at the time it started to leave sleid martes of the particular length and type and under the circumstances laid down by plaintiff’s automobile. To impeach Laitone, defendant offered Currie’s testimony relating the length of sleid marks laid down after the driver determined to apply the brakes at a given speed. Thus, it did not take into consideration the element of reaction time and could only confuse rather than clarify the issue at hand. It served to distract *25 the attention of the jury by submitting to them a variety of collateral issues incidental to the experiment including, in addition to the question of reaction time, the subsequently discovered 15 percent error in the test car’s speedometer. Evidence of this type was held inadmissible because of the collateral issues involved in Garcia v. Hoffman, 212 Cal.App.2d 530, 536 [28 Cal.Rptr. 98].

Other grounds existed from which the trial court could properly conclude that the experimental evidence was confusing and not dependable. The witness Currie was the passenger in the test car. The testimony as to the braking power applied was pure hearsay. The driver, who was the attorney for defendant, did not testify and was not subject to cross-examination. Contrary to defendant's assertion, Dr. Laitone did not indicate that the circumstances of the test conducted by Currie and Britton were comparable to those of the accident, but indicated that the test was not a proper one because of the absence of an external observer to note the distance traveled between the time the driver moved to apply the brakes and the beginning of the skid marks.

Finally, defendant argues that even if the trial court erred in failing to instruct the jury as to the limited purpose for which the Currie experiment was admitted, the error was waived by plaintiff's failure to remind the trial court to give such an instruction. The record indicates that on ruling on the admissibility of the evidence for purposes of impeachment only, the trial court indicated it would give the jury an appropriate instruction at the termination of the arguments. Admittedly, the court failed to do so in its instructions to the jury. In some circumstances, failure to object can constitute a waiver. But the rules applicable to such a waiver, to invited error, or to estoppel, have no application when an appellate court is considering the propriety of an order granting a new trial. If the trial court had denied the new trial, such an error would be considered waived by failure to object. But the trial court has broad discretion in considering motions for a new trial. Here, the court had determined that there had been “a miscarriage of justice,” and definitely felt a new trial was required. The trial court’s discretion should be upheld even where the error was a minor or debatable one. There is no necessity of showing prejudice in such a ease. In the instant ease, even if the failure to give the instruction were only debatably erroneous, the trial court’s discretion in granting the new trial cannot be disturbed (Mal *26 kasian v. Irwin, 61 Cal.2d 738, 747 [40 Cal.Rptr. 78, 394 P.2d 822]).

As indicated above, the instant case, on its facts, was a very close one. All of the evidence concerning the speed of both automobiles was in conflict or ha'd to be deduced by the experts from the circumstantial evidence of the skid marks and the position of the cars after the accident. Defendant admittedly did not see the yield sign; plaintiff accelerated her speed before entering the intersection. Under these circumstances, the trial court properly concluded as a matter of law that the evidence of the Currie experiment was not sufficiently probative and that the error of admitting this evidence was compounded by its failure to give a limiting instruction. We conclude that the order granting the new trial was properly granted.

The order appealed from is affirmed.

Shoemaker, P. J., and Agee, J., concurred.