BECK, Judge:

This is an appeal from an order granting a new trial in a products liability action. Plaintiff/appellant Wilkes-Barre Iron & Wire Works’ plant sustained severe damage when a liquid propane cylinder supplied by defendant/appellee Par-gas of Wilkes-Barre exploded. While an employee of appellant was using the cylinder, it slipped and fell. The valve on the cylinder broke off in the fall, and propane gas leaking from the damaged cylinder caused an explosion.

*287 Appellant’s suit against Pargas sounded in both negligence and strict liability under section 402A of the Restatement (Second) of Torts (1965). The negligence claim was non-suited at the close of appellant’s case. The essence of appellant’s strict liability claim was that the cylinder was defectively designed because it lacked a protective collar around the valve which would have shielded the valve from the impact received in the fall.

The jury returned a verdict in favor of appellant, but the trial court granted appellee’s motion for a new trial, holding that crucial testimony of appellant's expert witness should have been excluded because it exceeded the “fair scope” of his pretrial report under Pa.R.C.P. No. 4003.5(c). This is the principal issue before us on appeal. For the reasons stated below we affirm the order granting a new trial.

The decision whether to grant or deny a new trial is within the sound discretion of the trial court. Carnicelli v. Bartram, 289 Pa.Super. 424, 433 A.2d 878 (1981). Although appellant cites us to a statement in Hilbert v. Katz, 309 Pa.Super. 466, 471, 455 A.2d 704, 706 (1983), that “an appellate court may be more exacting in reviewing a new trial grant than a new trial denial,” because the grant of a new trial interferes more with the function of the jury, the context of that statement was strictly with respect to new trials granted on the ground that the verdict was against the weight of the evidence. Elsewhere in the Hilbert opinion, we specifically state with respect to new trials awarded for other reasons, including as in the instant case erroneous rulings on the admission of evidence, the trial court has “broader discretion.” Therefore the appropriate standard is still whether the trial court has abused its discretion.

We hold that the trial court did not abuse its discretion in ruling that a new trial was required because the testimony of appellant’s expert witness exceeded the fair scope of his pretrial report. Discovery of information concerning expert testimony is governed by Pa.R.C.P. No. 4003.5. Subsection (a)(1)(b) of the Rule states in pertinent part:

*288 A party may through interrogatories require ... the other party to have each expert so identified by him state the substance of the facts and opinions to which the expert is expected to testify and a summary of the grounds for each opinion.

Subsection (c) of the Rule then defines the permissible scope of the expert’s trial testimony:

To the extent that the facts known or opinion held by an expert have been developed in discovery proceedings under subdivision (a)(1) or (2) of this rule, his direct testimony at the trial may not be inconsistent with or go beyond the fair scope of his testimony in the discovery proceedings as set forth in his deposition, answer to an interrogatory, separate report, or supplement thereto. However, he shall not be prevented from testifying as to facts or opinions on matters on which he has not been interrogated in the discovery proceedings.

In the instant case, appellant retained Mr. Frederic Blum, a mechanical engineer, to testify regarding the cause of the explosion with particular reference to the design of the propane tank. In response to interrogatories served by Pargas, appellant Wilkes-Barre Iron & Wire Works disclosed Mr. Blum’s identity and attached his report which read as follows:

National Fire Protection Association/ANSI Standard No. 58, concerning Storage and Handling of Liquified Petroleum Gases (1976) states in Paragraph 2130 that:

Portable containers of 1,000 pounds (nominal 120 gallons) water capacity or less shall incorporate protection against physical damage to container appurtenances and immediate connections to these while in transit, storage, while being moved into position for use, and when in use ... by (means of): (b) A ventilated cap or collar____ Construction (of the cap or collar) shall be such that the force of a blow will not be transmitted to the valve.

The subject tank had no such protective collar, although many similar tanks provided by Pargas did. If the sub *289 ject tank had been provided with such a collar, it is clear the accident would not have occurred.

The falling tank would have struck the object with the collar instead of the valve. The valve would have experienced only a low-acceleration force due to inertia instead of the high-acceleration shock of direct impact. As a result, the valve would not have broken off. It is just such accidents that protective collars are designed to prevent.

C. CONCLUSIONS

(1) The tank’s shut-off valve broke off due to impact with a foreign object, as reported.

(2) The accident would not have occurred if the tank had been fitted with a protective collar around its shut-off valve as required by NFPA/ANSI regulations. The lack of a collar represents a clear safety violation on the part of the supplier, Pargas.

Record at 71a, 72a, 76a (emphasis added).

Significantly, Mr. Blum’s report was silent as to the means for attaching the protective collar to the cylinder. The NFPA/ANSI standards cited do not state that the collar must be welded on. At trial, however, Blum specifically testified that the cylinder was defective because it lacked a permanently welded-on protective collar. Blum stated that a cylinder supplied only with a detachable collar would not be completely safe because the collar could be accidentally or deliberately removed, leaving the valve unprotected.

The trial court allowed this testimony over Pargas’ objection. Pargas argues that Blum’s trial testimony exceeded the fair scope of his report because the report did not put Pargas on notice that only a permanently attached collar would have been sufficient, in Blum’s opinion, to discharge Pargas’ duty to produce a safe product.

We agree. Both parties cite the same line of cases construing Rule 4003.5(c) and similar local rules, each side *290 claiming that the cases support its position. See, e.g., Augustine v. Delgado, 332 Pa.Super. 194, 481 A.2d 319 (1984); Klyman v. SEPTA, 331 Pa.Super. 172, 480 A.2d 299 (1984); Mapp v. Dube, 330 Pa.Super. 284, 479 A.2d 553 (1984); Martin v. Johns-Manville Corp., 322 Pa.Super. 348, 469 A.2d 655 (1983); Brogley v. Chambersburg Engineering Co., 306 Pa.Super. 316, 452 A.2d 743 (1982).

The experience of this Court as revealed in these cases has been that it is impossible to formulate a hard and fast rule for determining when a particular expert’s testimony exceeds the fair scope of his or her pretrial report. Rather, the determination must be made with reference to the particular facts and circumstances of each case. The controlling principle which must guide us is whether the purpose of Rule 4003.5 is being served. The purpose of requiring a party to disclose, at his adversary’s request, “the substance of the facts and opinions to which the expert is expected to testify” is to avoid unfair surprise by enabling the adversary to prepare a response to the expert testimony. See Augustine v. Delgado, 332 Pa.Super. at 199, 481 A.2d at 321 (“Pa.R.Civ.P. 4003.5 favors liberal discovery of expert witnesses and disfavors unfair and prejudicial surprise”); Martin v. Johns-Manville Corp., 322 Pa.Super. at 358, 469 A.2d at 659 (“[W]e have found experts’ reports to be adequate ... when the report provides sufficient notice of the expert’s theory to enable the opposing party to prepare a rebuttal witness.”)

In other words, in deciding whether an expert’s trial testimony is within the fair scope of his report, the accent is on the word “fair.” The question to be answered is whether, under the particular facts and circumstances of the case, the discrepancy between the expert’s pretrial report and his trial testimony is of a nature which would prevent the adversary from preparing a meaningful response, or which would mislead the adversary as to the nature of the appropriate response.

*291 In the instant case it is clear that appellee Pargas was prejudiced by the discrepancy. The fundamental issue in this products liability action is whether the propane cylinder was in a safe condition when it left appellee’s control. Mr. Blum’s report simply indicated that it was his opinion that a cylinder lacking a protective collar was defective. Appellee was therefore prepared to defend with evidence that the cylinder in question was equipped with a collar (albeit a removable one) when it was delivered to appellant. This defense was of course ineffectual against the claim that the only safe procedure was to weld the collar to the cylinder.

Appellant argues that the expert’s report was adequate because it disclosed the basic theory that the absence of a collar caused the accident, and the absence of a fixed collar is just a specific example included within the scope of the report. As a matter of pure logic, this argument has some appeal, but it misses the point. There were two courses of action open to Pargas to protect the valves on its cylinders. We shall call installing a removable collar Option A and installing a fixed collar Option B. Mr. Blum’s report states that a propane cylinder must have a collar conforming to certain standards to be safe. Pargas could choose either Option A or Option B and meet those standards. However, according to Blum’s trial testimony, appellee would have supplied a safe product only if it chose Option B. In other words, Blum did not simply state that a welded-on collar was a particular type of safe collar; rather, his trial testimony changed the parameters of what he deemed safe and unsafe. Appellee was thus seriously misled in the preparation of its defense. 1

The trial court correctly concluded that Blum’s testimony relating to the absence of a welded or permanent collar constituted a different and distinct defect on which liability *292 could be predicated. The trial court noted that this type of defect had not been previously articulated and was not addressed in the expert’s report. Blum had failed to indicate prior to trial that the cylinder was defective because its collar was not welded or permanently affixed to the cylinder. A fair reading of Blum’s report indicates that the only design defect contemplated by it was the absence of any protection to the shut-off valve in the event of a fall.

The expert’s opinion offered at trial was based on a design defect not in his report and one different in character from that described in his report. Blum’s testimony presented at trial permits the inference that appellee Par-gas supplied a defective cylinder because the company did not consider the possibility that a non-permanent collar might be removed or fall off. Under Blum’s in-court statement, Pargas would have supplied a defective product unless the collar were welded or permanently affixed. This new definition of design defect took appellee by surprise and constituted the kind of surprise that Rule 4003.5 seeks to prevent.

“Fair scope” contemplates a reasonable explanation and even an enlargement of the expert’s written words. However, it does not constitute a new and different definition of design defect.

The trial court noted that hindsight instructed it that Pargas had established prejudice and that the testimony complained of should have been excluded. At page 19 of its written opinion, the trial court observes that

[t]he record clearly reflects that Pargas was prepared to defend and did introduce evidence in the form of a receipt signed by an agent of Wilkes-Barre Iron indicating that the cylinder was delivered on February 17, 1977, with a protective collar, albeit a non-permanent one. Since that date was the one on which the product left the control of Pargas, it was the crucial point in time for determining whether the cylinder was indeed defective within the meaning of [Section] 402A. Pargas understandably then had reason to forego the use of an expert witness to rebut the opinions and grounds contained in Blum’s re *293 port. By establishing that such a protective collar did accompany the cylinder and that an employee of Wilkes-Barre Iron had removed the collar, the detrimental impact of Blum’s conclusions as contained in the report would have been absolutely negated. However, with the insertion of the new theory [new definition of defect], Pargas was surprisingly caught off guard without the availability of an expert to rebut that opinion testimony. It is also significant to note that the opinion complained of concerned one of the ultimate issues involved in this case. Accordingly, we are satisfied that Pargas was prejudiced when the court permitted the introduction at trial of what was, in effect, a new theory of liability [new definition of defect].

Appellant claims that any error was harmless for several reasons. First, appellant argues that other competent evidence on the absence of a welded collar was properly admitted at trial, thereby rendering the erroneous admission of such evidence harmless error. Appellant also asserts that this evidence was known to Pargas long before trial, thereby obviating any claim of surprise.

We disagree. The evidence referred to consists of deposition and trial testimony of a district manager of Pargas (called by appellant at trial as on cross-examination) that Pargas began fitting welded collars on its tanks after the accident on appellant’s premises occurred. 2 Although this is evidence that Pargas knew that the collars could be welded on and in fact did so subsequent to the accident, it does not address the question whether welding the collar on was necessary to make the tank safe. Therefore it is not duplicative of Blum’s expert testimony. As for the significance of the district manager’s substantially similar deposition testimony, we do not believe that the deposition of a third person can substitute for deficiencies in the report of an expert witness. A party who has received inadequate notice of the facts and opinions of an opponent’s expert *294 cannot be expected to prepare a response on the speculative supposition that his testimony will be consistent with that of some other witness.

Finally, appellant contends that the admission of Blum’s testimony was harmless error because the case was submitted to the jury on three theories of liability, only one of which depended on Blum’s opinion that the collar should have been welded on, and the jury returned only a general verdict.

The three theories of strict liability were: 1) that the cylinder was defective because it was delivered without a protective collar; 2) that even if delivered with a protective collar, the cylinder was nevertheless defective because the collar was not permanently affixed or welded to the cylinder, and 3) that Pargas supplied a defective product by refilling the tank with highly explosive propane.

Appellant urges us to hold that because the jury may have found in its favor without relying on Blum’s testimony, the jury’s verdict should be upheld. This contention is meritless.

In view of the general verdict, we cannot determine on what theory the jury’s decision was premised. Since the challenged testimony addressed one of the ultimate issues involved in the case, we can only speculate as to the basis for the jury’s verdict. It is precisely because we can only speculate as to the basis for the jury’s verdict that the verdict cannot be upheld. Where the verdict of the jury may have been based on improperly admitted evidence, the grant of a new trial is appropriate. Bell v. Western Pennsylvania Hospital, 293 Pa.Super. 37, 437 A.2d 978 (1981); Kelly v. Buckley, 280 Pa.Super. 353, 421 A.2d 759 (1980); Warren v. Mosites Construction Co., 253 Pa.Super. 395, 385 A.2d 397 (1978). Here the jury’s verdict clearly could have been based on the improperly admitted evidence. We agree with the trial court that under these circumstances a new trial is warranted to give both parties the opportunity to address all the issues fully and fairly.

The order of the trial court is affirmed.

Notes

. Contrary to appellant’s further suggestion, the statement in the report that the cylinder should have a collar protecting the valve does not necessarily imply that the collar must be welded on. There arc many examples in common knowledge of safety devices which are removable or defeatable by the user.

. Evidence of post-accident modifications is admissible in a products liability case. Matsko v. Harley-Davidson, 325 Pa.Super. 452, 473 A.2d 155 (1984).