JUSTICE GOLDENHERSH delivered the opinion of the court:

Plaintiff, Ronnie McKenzie, appeals from the denial of his post-trial motion for a new trial and his motion for leave to amend his complaint to conform the pleadings to the proof, by the circuit court of Williamson County. The underlying cause of action filed by plaintiff was a products liability suit based upon strict liability in tort against defendant, SK Hand Tool Corporation, for injuries plaintiff sustained when the ratchet wrench plaintiff was using broke while he was working on a truck, causing him to fall and injure himself. The case was tried to a jury, and at the close of all evidence, the jury returned a verdict in favor of defendant.

On appeal, plaintiff's principal contentions are that (1) the trial court erred in striking all evidence regarding defendant's wrench specifications, measurements of the component parts of the wrench in question, and plaintiff's expert witness' testimony with respect to those measurements and specifications and (2) the trial court erred in admitting evidence of an absence of prior accidents or incidents because defendant failed to establish the proper foundation for such testimony. We reverse and remand for a new trial.

I

Plaintiff was injured on September 10, 1987, while working on the engine of a truck at his place of employment, Uselton Sales Service (hereinafter Uselton) in Marion. On the day of the accident, plaintiff was to overhaul the engine on a large over-the-road truck. To overhaul the engine, plaintiff was required to remove the heads, which were held to the engine by a number of large bolts. Normally plaintiff used a 3/4-inch air impact wrench to remove the bolts; however, after removing several bolts, the wrench developed a low-pressure problem. Plaintiff decided to use the 3/4-inch ratchet wrench his employer had in the shop. This 3/4-inch wrench was manufactured by defendant. Plaintiff broke two or three bolts loose with the ratchet wrench and removed those bolts with the wrench. Plaintiff, in attempting to remove another bolt, experienced difficulty even though the bolt broke loose. To remove the bolt, plaintiff put an extension on the wrench. While standing on the truck, plaintiff began pulling back the wrench. The wrench came apart and plaintiff fell off the truck backwards, hitting his right side and shoulder on the concrete floor. Plaintiff sustained injuries to his neck and shoulder.

When he got up, plaintiff found pieces of the wrench on top of the engine, in the frame of the truck, and on the floor. Plaintiff picked up all the pieces of the wrench and put them on top of the battery box. Plaintiff did not find the plug from the wrench. Plaintiff was alone at Uselton when the accident occurred. When David Couty, plaintiff's co-worker, and Gene Uselton, the shop owner, returned to the shop, plaintiff told them about the accident and showed them the wrench.

Gene Uselton testified that the 3/4-inch-drive ratchet wrench was not used often in his shop and that he could not remember when the wrench was purchased. He did not know whether anyone had ever disassembled the wrench since he got it. Plaintiff also stated that to his knowledge no one had ever disassembled the wrench. Plaintiff testified that he had used this wrench before without any problems prior to the occurrence.

Len O'Connell, a manager with defendant, testified that the wrench in question was a 3/4-inch-drive ratchet wrench manufactured by defendant. O'Connell also testified that defendant made three other different-sized ratchet wrenches: a 1/4-inch drive, a 3/8-inch drive, and a 1/2-inch drive. The same design is used for all of these wrenches. However, some components of the inner body of the wrench are manufactured by other companies and purchased by defendant for use in assembly of the ratchet wrench. Defendant only makes the drive body, the pawl, the handle, and the reversing stem.

The inner body, or the driver, of the wrench in question fits into the housing of the handle. These parts are held together by a circular device called a snap ring. When the tips, or "ears," of the snap ring are pinched together, the driver can be removed from the handle. The snap ring holds the driver and the handle together. The spring in the ring is designed to insure the snug fit of the ring in the grooves.

Defendant has blueprints that contain the specifications of the sizes of each component of the wrench. The components are to comply with these specifications, which have a tolerance for each measurement. The part is acceptable if its actual measurement fits within the tolerance limits. For each part there are figures that represent its upper and lower limits. If the measurement of the part does not fall between the upper and lower limits, the machinist knows the part is not acceptable for use.

The component parts of the wrench in question were measured by four individuals, and the measurements were compared to those required by the specifications. Dr. Leon Dunning, a retired professor of mechanical engineering, and William Switalski, a mechanical engineer, measured the parts and testified as expert witnesses for defendant. Charles Reynolds, a registered mechanical engineer, and Hurley Johnson, chief inspector for a parts manufacturer, measured the component parts and testified on behalf of plaintiff. The outside diameter of the snap ring and its "ears" were measured when compressed at .56 and .59 inches apart, and the degree of hardness of the ring was determined using a Rockwell C scale. Also, the outside diameter of the snap ring groove in the handle and the outside and inside diameter of the snap ring slot in the driver were measured and compared to defendant's specifications. Plaintiff's expert, Reynolds, noticed a tapering radius on both the inner and outer edges of the snap ring groove in the handle. This taper was also measured.

Plaintiff's experts measured the hardness of the snap ring several times on a Rockwell C scale and got several different hardnesses ranging from 45 to 51 when the ring was measured in various places. Reynolds concluded that the ring did not meet the hardness required in the specifications (a measurement of 48 to 52 on a Rockwell C scale) and that this failure to comply with the specifications could affect the snug fit of the ring in the grooves of the driver and thus its ability to properly hold the driver and the handle together. Reynolds also stated that the variance in hardness found in the ring was the result of manufacturing and not use. Defendant's experts did not measure the snap ring for hardness.

Upon measuring the "ears" of the snap ring compressed at .56 and .59 inches apart, plaintiff's expert got a diameter of 2.269 inches and 2.280 inches, respectively. Plaintiff's expert found the measurement of the outside diameter of the snap ring groove in the handle to be 2.3130 to 2.3125 inches. Defendant's expert's measurement for the same part was 2.315 inches. Defendant's specifications required 2.290 inches with a tolerance of .005 inches, so a measurement between 2.285 inches and 2.295 inches would be acceptable. However, the measurements obtained by both plaintiff's and defendant's experts exceeded defendant's specifications.

Reynolds, one of plaintiff's experts, noticed a tapering in the radius of both the inside and outside edges of the snap ring groove in the handle. Defendant's specifications do not mention tapering in the radius on the edge of the ring. Reynolds testified that it is important that the internal parts of the wrench comply with the specifications, particularly with regard to the snap ring and the grooves. Reynolds stated that it is common knowledge in the industry that the dimensions of a groove which retains a snap ring must be perpendicular and parallel and that the edges must remain sharp. Reynolds opined that a taper is like a ramp which allows the snap ring to be ejected. Also, in his opinion, the tapering was done during manufacturing and did not result from use. This point was affirmed by David Bidlack, engineering manager of defendant's Ohio plant. Bidlack stated that the taper is intended to prevent the cutting machine from dragging on the part as it comes back out of the groove. Hurley Johnson and Leon Dunning also noticed the tapering.

Regarding the possible causes of the accident, Reynolds offered two potential explanations for why the wrench came apart. First, Reynolds opined that the failure of the parts to comply with defendant's specifications would create a manufacturing defect which could contribute to an incident of the sort involved here. Specifically, the snap ring groove is not to specification, and the handle has a large radius and tapering of the edges. Reynolds suggests, as a second possible cause of the accident, improper reassembly of the snap ring either during servicing of the wrench or during assembly by the manufacturer.

At the end of direct examination, Reynolds testified that in his opinion, the wrench in question was defective and unreasonably dangerous because the snap ring groove did not comply with defendant's specifications and, in conjunction with the outward taper of the snap ring groove, this overall noncompliance with the specifications could permit the wrench to come apart.

Upon the close of plaintiff's case, defendant's motion for directed verdict was denied. At the close of all the evidence, plaintiff filed his motion for leave to amend his complaint, which the court denied. Also, defendant filed its motion for directed verdict. After the conference on instructions and before closing arguments, the court instructed the jury that it had granted motions to strike the testimony of plaintiff's and defendant's experts "with respect to measurements and specifications with respect to the effect of those specifications and measurements on the wrench." The jury returned a verdict in favor of defendant and against plaintiff. The court denied plaintiff's post-trial motion, from which plaintiff appeals.

II

Plaintiff contends that the trial court erroneously struck all evidence relating to the measurements and specifications of the component parts of the wrench and plaintiff's expert's testimony with respect to those measurements and specifications. Furthermore, plaintiff argues that the trial court erroneously relied on D'Olier v. General Motors Corp. (1986), 145 Ill. App.3d 543, 495 N.E.2d 1040, in striking the measurement and specifications evidence and plaintiff's expert's testimony because this evidence and testimony referred to the cause of plaintiff's injury and not the defect in the wrench. Plaintiff maintains that the existence of a manufacturing defect was already proven because the measurements of the parts of the wrench were shown not to comply with the manufacturer's specifications. We agree.

To prevail in a strict liability case, a plaintiff must prove that his injuries resulted from an unreasonably dangerous or defective condition of the product and that the condition existed at the time the product left the manufacturer's control. ( Tweedy v. Wright Ford Sales, Inc. (1976), 64 Ill.2d 570, 357 N.E.2d 449; McColgan v. Environmental Control Systems, Inc. (1991), 212 Ill. App.3d 696, 699, 571 N.E.2d 815, 817.) Strict liability applies if the product is found to be "unreasonably dangerous when it fails to perform in the manner reasonably expected in light of its nature and function." ( McColgan, 212 Ill. App.3d at 699, 571 N.E.2d at 817.) A product may be found to be unreasonably dangerous by virtue of a design or manufacture defect. ( Renfro v. Allied Industrial Equipment Corp. (1987), 155 Ill. App.3d 140, 155, 507 N.E.2d 1213, 1226.) Under strict liability, a plaintiff need not prove a specific defect. ( Varady v. Guardian Co. (1987), 153 Ill. App.3d 1062, 1066-67, 506 N.E.2d 708, 711.) A defect may be proven inferentially either by direct or circumstantial evidence. ( Varady, 153 Ill. App.3d at 1067, 506 N.E.2d at 711.) The circumstantial evidence must justify an inference of probability, as distinguished from mere possibility, because liability cannot be predicated on mere speculation, guess, or conjecture. ( Varady, 153 Ill. App.3d at 1067, 506 N.E.2d at 711.) Furthermore, the issue of whether a product is defective is a question of fact for the jury. ( Renfro, 155 Ill. App.3d at 155, 507 N.E.2d at 1226.) Thus, the issue of whether a defendant's failure to meet its own design specifications constitutes a defective condition that was unreasonably dangerous to a plaintiff is a question of fact and is properly presented to the jury. Fink v. Chrysler Motors Corp. (1974), 16 Ill. App.3d 886, 890, 308 N.E.2d 838, 841.

In the present case, the evidence is sufficient to establish the presence of a defect in the wrench. The wrench in question has a number of component parts, the inner body, or driver, and handle which are held together by a snap ring that fits into a groove in the handle and driver. Defendant has design specifications which set forth measurements with which the parts must comply. Each part has upper and lower limits within which a part must fall. If the part does not fall within these limits, the machinist knows the part is not acceptable for use. Plaintiff's experts, Reynolds and Johnson, measured these component parts. These experts found that the hardness of the snap ring was not uniform. When measured in various places, the hardness of the ring ranged from 45 to 51 on a Rockwell C scale. Defendant's specifications require a measurement of 48 to 52 on the Rockwell C scale. The outside diameter of the snap ring groove in the handle measured 2.3130 to 2.3125. The specifications required 2.290 inches, with a tolerance of .005 inches, so a measurement between 2.285 to 2.295 inches would be acceptable. Therefore, the outside diameter of the snap ring groove in the handle is larger than the diameter required in the specifications. Plaintiff's expert also found a tapering in the radius of both the inside and outside edges of the snap ring groove in the handle. The design specifications do not require a tapering in the radius of either edge of the ring. Reynolds said, in explaining the importance of the internal parts' compliance with the design specifications, that the grooved dimensions must be perpendicular and parallel and the edges must remain sharp. When the edges are not sharp, the axial forces will cause the ring to pop out of the groove. Reynolds said that "with the taper, it is like pushing something up a ramp," thereby allowing the snap ring to eject from the handle. Moreover, Reynolds testified in direct examination that, in his opinion, the wrench was defective and unreasonably dangerous because it did not comply with the specifications, thereby creating the possibility of the wrench coming apart while being used. Furthermore, testimony by plaintiff and his co-workers indicates that the wrench had not been disassembled since its purchase, thereby eliminating a possible secondary cause. Plaintiff has established a prima facie case that the wrench was defective, that the defect existed when it left defendant's control, and that in the absence of abnormal use or secondary causes, the wrench failed to perform in the manner reasonably expected in light of its nature and intended function. Varady v. Guardian Co. (1987), 153 Ill. App.3d 1062, 1066, 506 N.E.2d 708, 711.

Defendant makes much of plaintiff's expert's inability to replicate the occurrence. Although plaintiff's expert was not able to replicate the occurrence or isolate a precise cause of the occurrence, absence of such proof is not fatal to plaintiff's case since plaintiff is not required to prove a specific defect. Tweedy v. Wright Ford Sales, Inc. (1976), 64 Ill.2d 570, 357 N.E.2d 449; Renfro, 155 Ill. App.3d at 156, 507 N.E.2d at 1227.

III

Defendant maintains that the trial court properly struck Reynolds' testimony because Reynolds' opinion amounted to speculation since he expressed a mere possibility, not a probability, that the part's noncompliance with defendant's design specifications would allow the wrench to come apart. Plaintiff argues to the contrary, stating that Reynolds' use of the word "possibility" as opposed to "probability" was made in connection to the issue of causation and not in conjunction with defect. We agree with plaintiff's contention.

An expert's opinion as to the cause of an injury is not improper or inadmissible because it is couched in terms of probabilities or possibilities based upon certain assumed facts. ( Rodrian v. Seiber (1990), 194 Ill. App.3d 504, 507, 551 N.E.2d 772, 774.) An expert may testify as to what might be the cause of the injury despite any objection that his testimony is inconclusive or speculative. ( Presswood v. Morris (1979), 70 Ill. App.3d 513, 517, 388 N.E.2d 844, 848.) Counsel may draw the jury's attention to possible deficiencies in the expert's inferences and the information the expert relied upon. ( Fullreide v. Midstates Beverage Co. (1979), 70 Ill. App.3d 758, 762, 388 N.E.2d 1070, 1073.) Most importantly, the jury, once aware of any "infirmities" in the expert's opinion, can determine his credibility. Rodrian, 194 Ill. App.3d at 507, 551 N.E.2d at 774.

The source of the controversy between the parties is a statement made by Reynolds during direct examination when Reynolds was asked why he believed the wrench was defective and unreasonably dangerous. Reynolds stated that he believed the wrench was "defective and unreasonably dangerous because it did not comply with the specifications, and that noncompliance with the specifications created the possibility of it coming apart while being used, due to the fact that the snap ring was not in compliance and the outward taper of the snap ring groove could have permitted it to come apart."

The context in which the word "possibility" appears is in connection with that part of the sentence referring to causation, not defect. An expert, when referring to the causes of an occurrence, may couch his answer in terms of "possibility." Use of the word "possibility" in conjunction with causation does not automatically render the expert's answer inadmissible or improper. ( Presswood, 70 Ill. App.3d at 517, 388 N.E.2d at 848.) Nor is Reynolds' testimony made speculative or inconclusive merely because he used the word "possibility" in connection with causation. ( Presswood, 70 Ill. App.3d at 517, 388 N.E.2d at 848.) Because the expert's testimony is but " 'the opinion of the witness, given, * * * [on] facts assumed to be true,' " the trier of facts still must determine the facts. Presswood, 70 Ill. App.3d at 517, 388 N.E.2d at 848, quoting Chicago Union Traction Co. v. Roberts (1907), 229 Ill. 481, 484, 82 N.E. 401, 402.

Defendant cited and the trial court relied on D'Olier v. General Motors Corp. (1986), 145 Ill. App.3d 543, 495 N.E.2d 1040, in striking all evidence relating to the measurements and specifications of the component parts, as well as the testimony of plaintiff's expert witnesses concerning the noncompliance of the component parts with defendant's design specifications. The reliance on D'Olier is misplaced.

In D'Olier, an injured motorist brought strict liability and negligence claims against the car dealer and manufacturer when she sustained injuries in a single-vehicle accident in which the steering wheel of the plaintiff's auto locked, leaving her unable to steer. In her strict liability claim, the plaintiff alleged that the steering mechanism of the vehicle was defective and in an unreasonably dangerous condition. For the 2 1/2 years prior to the accident, the plaintiff did not have any mechanical problems with her car and did not notice any signs of potential mechanical difficulties. At trial, the plaintiff's expert witness testified to three "possible" causes of the alleged steering malfunction. The expert raised the possibility that an O-ring on a fitting of the power steering gearbox may have entered the system, that a brass tab found on the same fitting might have entered the system, and that the steering fluid could have become contaminated prior to the accident. On cross-examination, the plaintiff's expert conceded that it was unlikely the O-ring was placed in the system during the manufacturing process, that he failed to find fragments of the tab in the steering gear and pump, and that the steering fluid was more likely contaminated by the elements subsequent to the accident, when the car was left outside at a storage yard with the hood partially raised. The trial court directed a verdict in the defendants' favor.

In affirming the trial court, the D'Olier court cited Belleville National Savings Bank v. General Motors Corp. (1974), 20 Ill. App.3d 707, 313 N.E.2d 631, for the proposition that the plaintiff is required to establish a probability, not possibility, that a defect existed before the automobile left the defendant's control. They further determined that the plaintiff had not met even the possibility standard. As to causation, the court in Belleville National Savings Bank phrased the issue as, "[D]id the plaintiff introduce any evidence from which it might reasonably be inferred that a condition of the product proximately caused plaintiff's decedent's demise?" Belleville National Savings Bank, 20 Ill. App.3d at 709, 313 N.E.2d at 633.

D'Olier is readily distinguishable from the present case. Here, the defect alleged by plaintiff certainly existed before the wrench left defendant's control. Both plaintiff's and defendant's experts found that the component parts did not meet defendant's design specifications. Defendant uses these specifications to determine the suitability of a part for use in the wrench. Bidlack, an engineer for defendant, testified that defendant randomly tests to determine whether parts meet defendant's specifications. Bidlack also testified that if a snap ring failed to meet specifications, he would "pull that particular ring aside." Further, plaintiff here presented evidence that neither plaintiff nor his co-workers disassembled the wrench prior to the occurrence. Thus, plaintiff showed that the wrench was defective when it left defendant's control.

Expert testimony on the measurements and specifications of the component parts and the part's failure to comply with defendant's specifications would be instructive to the jury in determining the issue of whether the wrench is defective and whether the defect is the proximate cause of plaintiff's injury. Thus, such expert testimony is properly presented to the jury. ( Fink, 16 Ill. App.3d at 890, 308 N.E.2d at 841.) In light of our review of the record, we conclude that the trial court erred in striking Reynolds' testimony regarding the measurements of the component parts and their noncompliance with defendant's design specifications, as well as other evidence with respect to the measurements and specifications.

IV

Plaintiff next contends that the trial court improperly admitted evidence regarding the lack of prior accidents because a sufficient foundation for such testimony was never laid. We agree.

Existing case law establishes that a party may introduce evidence of the lack of prior accidents or incidents when that party establishes as a foundation that (1) the same product was used and (2) the product was used under conditions substantially similar to those in which plaintiff used the product. ( Balsley v. Raymond Corp. (1992), 232 Ill. App.3d 1028, 1030, 600 N.E.2d 424, 425.) The decision to admit or refuse to admit evidence is within the trial court's discretion, and its decision should be reversed on appeal only if the court has abused its discretion. Balsley, 232 Ill. App.3d at 1032, 600 N.E.2d at 427.

In the instant case, defendant's witness, a corporate personnel manager for defendant, testified that defendant made about 7,000 of the 3/4-inch-drive ratchet wrenches each year. Defendant also manufactures approximately 350,000 smaller wrenches annually. The witness testified that he reviewed existing corporate records and found that no claim, return, allegation, or anything else indicating problems with the 3/4-inch ratchet wrenches or any of the various other wrenches manufactured by defendant had ever been made. He further testified that no claims had been made with respect to the snap ring dislodging while the wrench was in use. The trial court admitted this evidence over objections from plaintiff's counsel.

In reviewing the testimony of defendant's personnel manager, it is clear that the evidence presented is without a minimal foundation. Defendant failed to establish that the same product was used. There was no evidence whether the other 3/4-inch-drive ratchet wrenches complied or failed to comply with defendant's design specifications. Consequently, no evidence was presented regarding prior accidents and incidents involving nonconforming wrenches. Further, evidence with respect to prior accidents and incidents involving other wrenches is irrelevant to the functioning of the 3/4-inch wrench involved here. Admission of this evidence was unduly prejudicial to plaintiff because its interjection permitted the jury to make unsupported inferences regarding the wrench in question.

Similarly, defendant's witness failed to present any evidence indicating that the absence of claims and complaints occurred when others were using the 3/4-inch wrench under conditions substantially similar to those used by plaintiff ( i.e., removing head bolts from a truck engine, in a crouching position four feet above the floor while using an extension and socket).

Defendant relies on Leischner v. Deere Co. (1984), 127 Ill. App.3d 175, 468 N.E.2d 182, to support its contention that the trial court properly admitted the evidence of a lack of prior accidents and incidents involving defendant's ratchet wrenches with a snap ring design. Defendant's reliance on this authority is misplaced.

In Leischner, the plaintiff brought a products liability action against a snowmobile manufacturer for a spinal cord injury sustained in a snowmobile accident. The plaintiff alleged that the seat and suspension system of the snowmobile were defectively designed. At trial, the manufacturer presented evidence that there were 64,000 snowmobiles in use. The manufacturer's expert then testified that he reviewed the manufacturer's snowmobile accident records which revealed that, except for the plaintiff's claim, there were no complaints of spinal injury caused by the seat and suspension system. The Leischner court concluded that the evidence established that a substantial number of the snowmobiles were used on similar terrain at similar speeds in a similar manner; thereby, the manufacturer established a proper foundation for lack-of-prior-accident evidence. Leischner, 127 Ill. App.3d at 177, 468 N.E.2d at 183.

Leischner is easily distinguished from the case at bar. Unlike the snowmobile in Leischner, which did not contain any variables that diminished substantial similarity in use, the nature of the product (a ratchet wrench) in the instant case permits the user to vary the conditions under which the wrench is used. Here, plaintiff was using the wrench to loosen head bolts on a truck engine from a crouching position four feet off the floor with an extension and socket added. Other variables that diminish substantial similarity in the use of the wrench include: (1) the compliance of the component parts with the design specifications, (2) the amount of force exerted and the manner in which it is applied by the user, (3) the use of other tools in addition to the wrench, and (4) the use of extensions on the wrench. Because defendant's expert did not establish that the same wrench was used and that the wrench was used in a substantially similar manner, defendant failed to meet the foundational requirements for the admission of lack-of-prior-accidents evidence. Accordingly, we find that such testimony was improperly admitted into evidence.

For the foregoing reasons, the judgment of the circuit court of Williamson County is reversed, and the cause is remanded for a new trial.

Reversed and remanded.

MAAG, P.J., concurs.