*329ORDER
CARROLL, United States Magistrate Judge.Plaintiffs’ decedent died as a result of injuries sustained when the 1985 Chevrolet S10 pickup with extended cab she was driving collided with the rear of a van. This is a design-defect case in which plaintiffs allege that the lock mechanism of the seat-belt failed to function, leaving the driver without the protection of the restraint system when the S10 made impact. The case has been rife with discovery disputes. On August 2, 1991, this court entered a protective order prompting defendant General Motors to be more forthcoming in discovery. On September 19, 1991, a hearing was held on various pending motions related to discovery. Among these discovery issues was the question of which restraint systems are “similar to” other systems. Plaintiffs contend that the allegedly defective device is generic; defendant contends that the device interacts with its environment to such a great extent that its performance in one vehicle line is absolutely nontransferable to another. Also at issue •is how far back plaintiffs should be allowed to reach in discovery. Plaintiffs contend for 1974, but defendants have been willing to produce only from 1983 forward. Finally, related to all of the requested records is defendant’s contention that production would be extremely burdensome, time-consuming, and expensive.
The “Similar” Vehicles or Systems Issue
What plaintiffs asked for in their First Interrogatories was information about “that certain seatbelt assembly or safety restraint system in the 1985 Chevrolet Pickup Truck, Model No. C10653 bearing VIN1GCCS14B1F2160927, and any other similar vehicle that uses the same safety restraint system or seatbelt assembly described in the complaint in this action.” The request is inartful, and a virtual taunt to engineering experts to prove why any system other than that in the actual crash vehicle is not the “same” or why any other vehicle type is not “similar.” In fairness to defendant, plaintiff did ask for information as to “similar vehicles,” which can be construed more readily (and perhaps more wilfully) by engineers than by laypeople. But plaintiffs’ counsel are not engineers. They would not know, until General Motors told them, that the restraint system at issue was a “body-mounted, dual spool, three-point manual belt system” or that it is a “single-retractor system.” Engineering nuances are not the ordinary province of most lawyers. General Motors seized the opportunity for evasion afforded by this query, expressed as it was in the terms an engineering novice would use: it limited its response to plaintiffs’ interrogatories to information about the driver’s seatbelt assembly on the 1985 Chevrolet S10 pickup truck with extended cab. In supplemental responses, General Motors also narrowly construed “similarity” of restraint systems and vehicle types.
In response to Plaintiffs’ Second Request for Production, General Motors again unilaterally limited its production to restraint systems manufactured from 1983 forward that it essayed the “same” or “similar” to the crash vehicle. After the court entered its August 2 protective order, General Motors broadened the compass of similarity it had previously drawn. It provided information about seven other vehicles with similar body types or sizes, excluding ordinary passenger cars such as sedans and coupes. General Motors also limited production of documents to those concerning model years 1983 forward in which a driver was injured in a frontal, nonrollover collision in which the retractor allegedly failed to lock as it was designed to do.
At the September 19 hearing, the court heard expert testimony from Mr. David Peruski, who has extensive experience with the company’s belt-restraint systems. Mr. Peruski testified that restraint systems are significantly different from car line to car line. The most obvious difference is the “geometry,” or the fit and placement within the car’s interior. The location and mounting of the restraint system differs among the various car lines. Moreover, Mr. Peruski testified that differently configured restraint systems are put into different models because the “crash pulse” of *330different vehicle lines will not be the same. A crash pulse in, e.g., a Cadillac will be different than that of a General Motors-manufactured van. In the event of a crash, the Cadillac driver would generally have a better chance of survival because the Cadillac provides more “crush space,” or more room for denting in between the engine components and the driver. The Cadillac will take longer to absorb the full impact and to react structurally. Greater crush space means buying a little more time for the occupant.
The court has no doubt that this is true. The testimony of Mr. Peruski was informative and, in another type of case, might have been compelling. However, this is a design-defect ease. The allegation is that this restraint system did not function properly because the lock mechanism did not do its job. This is not a case about the relative safety afforded by different body types or about the crashworthiness of certain types of auto bodies. It is also not about what type of restraint system best fits the geometry of a given vehicle. The “crash pulse” testimony would be seductive in another type of case, but it is not here. Here, the issue is whether that lock mechanism, a swinging pendulum design, functioned properly according to a proper design. The allegation is that it did not. The undisputed fact is that the swinging pendulum design is the same locking device used in all General Motors vehicles, from the Cadillac to the S10. The mechanics of the seatbelt’s function are the same for a Cadillac as for an S10, as Mr. Peruski conceded on cross-examination. The concept is the same, and the pendulum-induced lockup is the same. The engineering nomenclature, geometric configuration within the vehicle, housing and placement of the restraint system, and a dozen other “crash pulse” factors will not change that essential fact.
This court cannot allow General Motors to define the parameters and content of discovery in this ease. The court believes that the issue of similarity or dissimilarity has become something of a red herring that has allowed General Motors to do just this. Therefore, the court takes this definitional matter out of the hands of the parties. Rule 26(b) provides that the parties may discover “any matter, not privileged, which is relevant to the subject matter involved in the pending action.” Information about similar products or devices is generally discoverable. See Kramer v. Boeing Co., 126 F.R.D. 690, 693 (D.Minn. 1989); Culligan v. Yamaha Motor Corp., 110 F.R.D. 122 (S.D.N.Y.1986). It is also admissible at trial. See Hessen v. Jaguar Cars, Inc., 915 F.2d 641 (11th Cir.1990); Jackson v. Firestone Tire & Rubber Co., 788 F.2d 1070 (5th Cir.1986); General Motors v. Van Marter, 447 So.2d 1291 (Ala. 1984). Here, we are not dealing with mere similarity, but rather, with virtual sameness. A restraint system built upon the essential component of a pendulum-induced locking mechanism is at issue here. It is alleged that the locking mechanism failed. Therefore, all systems dependent upon a pendulum-induced locking mechanism have relevance to this litigation; all General Motors vehicles use this same locking mechanism. The concept is the same, and the design is reasonably generic. Plaintiffs therefore are entitled to discover information beyond that provided about the several vehicle lines which General Motors contends have a “similarity” to the S10. How much more they should discover is, not surprisingly, also a matter of contention.
Cost and Burden of Producing All Requested Data
General Motors’ objection to broader discovery is phrased in terms of burden and expense. At the September 19 hearing, General Motors put forward Mr. Joe Ziembo, Director of the Product Information Group. Mr. Ziembo handles all discovery requests for the company. Mr. Ziembo testified that, were the court to order production of data pertaining to all vehicles since 1970, the crash test results alone would number 4,300. (There are some 15,-300 sled tests.) Each test consumes 300 feet of 16-millimeter film and is accompanied by 16 still photos and 20 pages documenting the test findings. Mr. Ziembo estimated that 1,000 personnel hours would be expended in determining if crash tests *331were responsive to plaintiffs’ needs, and 1,200 hours would be necessary for sled tests (15 minutes viewing time per test). This estimate does not include the custodial time tracking them down from each division; this is only what Mr. Ziembo’s group would exhaust.
The burden is ever more draconian, according to Mr. Ziembo, when costs are considered. Most of these tests are now on fiche, and Mr. Ziembo testified that the conversion process was lengthy as well as expensive. Producing the crash tests only, dating back to 1970, and making only one copy of each, would cost $180,000. Mr. Ziembo testified that three copies would be required to satisfy plaintiffs’ discovery request, bringing the cost to $540,000. This figure is for crash tests only and does not include sled tests. Mr. Ziembo testified that it would take one year to produce all of this.
In response to plaintiffs’ query of what General Motors had produced in previous litigation, Mr. Ziembo conceded that there were some research shortcuts that could be used. For instance, a search can be restricted using computer features, but Mr. Ziembo says that his division lacks the capability to perform, e.g., a subset search for dual versus single-belt system failures. Mr. Ziembo did testify, however, that a centralized location exists for CPIRs (accident reports), and that these reports are indexed in some fashion. Mr. Ziembo was not sure that the index was such that it could be used to locate all restraint-system failures.
The law applicable to an objection to production on grounds of burdensomeness and expense is fairly clear. The mere fact that producing documents would be burdensome and expensive and would interfere with a party’s normal operations is not inherently a reason to refuse an otherwise legitimate discovery request. Biliske v. American Live Stock Ins. Co., 73 F.R.D. 124 (W.D.Okla.1977); Keco Indus., Inc. v. Stearns Elec. Corp., 285 F.Supp. 912 (E.D.Wis.1968); Speedrack, Inc. v. Baybarz, 45 F.R.D. 254 (E.D.Cal.1968); Technograph, Inc. v. Texas Instruments, Inc., 43 F.R.D. 416 (S.D.N.Y.1967); Rockaway Pix Theatre, Inc. v. Metro-Goldwyn-Mayer, Inc., 36 F.R.D. 15 (E.D.N.Y.1964). Scholarly authority clearly perceives the relationship between the enormity of a corporation like General Motors and the paper shield it can erect to discovery. As one treatise has expressed it, “The fact that defendant’s size requires it to keep a great amount of records cannot give it immunity which a small organization would not possess.” 4A Moore’s Federal Practice § 34.19 n. 10.
Nor can the lack of an adequate filing system insulate a party from discovery. For example, in Baxter v. Travenol Labs., Inc. v. LeMay, 93 F.R.D. 379 (S.D.Ohio 1981), plaintiff opposed discovery on the grounds that producing 800,000 sales invoices would require it to search almost 3,000,000 documents. The cost, it was asserted, would be some $80,000, and “hundreds of man-hours.” The court ruled that an unwieldy recordkeeping system could not be used to frustrate discovery. Accord Alliance to End Repression v. Rochford, 75 F.R.D. 441 (N.D.Ill.1977). Similarly, in Kozlowski v. Sears Roebuck & Co., 73 F.R.D. 73 (D.Mass.1976), the court found that a corporation that maintained its records in such a way as to make document retrieval totally impracticable could not use this impracticability as grounds to defend against production. Accordingly, when plaintiff sought disclosure of complaints similar to her own against the company, the company was not permitted to interpose a defense to production that the indexing of the documents was not conducive to a subject-matter search.
Nevertheless, there is some authority for the proposition that, when the volume of material sought would make copying and shipment difficult and oppressive for the party in possession, or when the distance between the parties is great, the court may demur from ordering production and may instead order inspection in a manner convenient to the party in possession. 4A Moore’s Federal Practice § 34.19 (citing Compagnie des Bauxites de Guinea v. Insurance Co. of North America, 651 F.2d 877 (3d Cir.1981)).
*332The court thinks it highly probable in this case that the plaintiffs could find what they need in the CPIRs. Testimony at the hearing disclosed that these CPIR files are indexed in some fashion. To avoid some of the expense of time and money attending the fiche records of the crash tests, the court orders that plaintiffs first be permitted to inspect, at the facility where they are kept, the CPIR records. Should plaintiffs require further discovery after this inspection has taken place, they may be permitted to make a similar inspection of the fiche records. Plaintiffs should clearly understand that the availability of this avenue of discovery does not give them a license for obstinacy or willful blindness as regards their inspection of the CPIR documents; the court orders this because it views this as a lower-cost, less time-consuming process than inspecting the fiche records, and not as an appetizer to the fiche records. Defendant, for its part, should clearly understand that, should the court later be persuaded that the additional discovery of the fiche records is warranted, the court will consider the case law cited supra as to recordkeeping that is conducive to inaccessibility in discovery. The court is mindful, and defendant should be mindful, that “where defendant’s record-keeping system was designed in such a way that defendant itself was forced to use special equipment to photocopy its records before it could read them, the defendant was required to bear the expense of putting the records into readable form for discovery; in such a case, where the extra cost of discovery was a product of defendant’s own recordkeeping system, the court would not shift the burden of the cost to plaintiff.” 4A Moore's Federal Practice § 34.19 (citing Delozier v. First Nat’l Bank of Gatlinburg, 109 F.R.D. 161 (E.D.Tenn.1986)). The court thinks that general proposition apposite to the instant litigation.
Limitation on Time Span Encompassed in Discovery
Finally, plaintiffs and defendant differ as to how far back plaintiffs should be permitted to delve into the records of General Motors. Plaintiffs want to go back to 1974, the year the restraint system used in the subject vehicle made its debut. Defendant is willing to surrender data only from 1983 forward, the year in which the system was installed in the S10. The court has already resolved the issues of “similarity” and sameness and has found that General Motors’ construction of those terms will not govern in this case. The year 1983 is thus not a relevant year. The court will instead order that data from 1980 forward be discoverable by the plaintiffs, because 1980 was the year in which the system at issue had been utilized or installed in virtually all General Motors vehicles. The court believes that this limitation will gain the most relevant data without prejudicing plaintiffs or overburdening defendant.
CONCLUSION
For the foregoing reasons, it is ORDERED that the defendant provide the plaintiffs, on or before January 6, 1992 or at any time agreed to by the parties, with access to the CPIR records of all accidents involving allegations that a pendulum-induced locking mechanism failed to operate properly, in any vehicle manufactured by General Motors and using that device in its restraint system during the years 1980 to the present. Defendant’s objections to production are OVERRULED, and plaintiffs’ alternative motion for a protective order is DENIED as moot.