RIPPLE, Circuit Judge.

Grace Cummins, an employee of Mullinix Packages, Inc. ("Mullinix"), severed three of her fingers while operating a trim press manufactured by Lyle Industries, Inc. ("Lyle"). She brought this products liability action against Lyle, alleging that the trim press was defectively designed and bore inadequate warnings. After conducting extensive hearings, the district court granted Lyle's motion in limine to exclude certain testimony by Ms. Cummins' expert witness, Dr. Thomas Carpenter, on the issues of the feasibility of alternative designs and warnings. The jury returned a verdict for Lyle. Ms. Cummins now appeals the district court's exclusion of Dr. Carpenter's testimony. For the reasons set forth in the following opinion, we affirm the judgment of the district court.

I. [2] BACKGROUND

Lyle Industries, Inc., a manufacturer of production machinery, sold an industrial trim press to Mullinix. When the trim press left Lyle's control, both sides of the press were equipped with safety-interlocked wire-mesh guard doors. If either of these side guard doors was opened or removed, the safety interlock circuit would stop the movement of the trim press and would disconnect power to the unit until the guard door was replaced. Lyle also installed a third, fixed guard beneath the cutting area. This guard, which did not have to be opened or removed to service the press or to change the cutting dies, was permanently bolted onto the trim press frame.

After purchasing the trim press, Mullinix elected to remove the fixed guard on the bottom of the trim press and to replace it with a scrap removal system of its own design. Mullinix designed the scrap chute to cover the opening created when the fixed guard was removed. However, the system incorporated a hinged door that permitted access to the cutting area of the trim press. Mullinix did not equip this scrap chute door with a safety interlock circuit.

Mullinix also designed the "Position Press for Auto Feed" ("PPAF") function on the trim press. After Mullinix designed the PPAF function, it showed its design to Lyle. Lyle then installed the PPAF control on the trim presses that it sold to Mullinix. The PPAF function stops the press in the "open" position; it is used by the operator during the initial set-up of the trim press. In order to use the PPAF function, the operator presses the "PPAF" control on the side of the trim press. When pressed, the "PPAF" control sends a signal to a limit switch on the fly wheel of the trim press. The limit switch causes an arm to make contact with a cam on the rotating fly wheel, which, in turn, causes a disc brake to be applied to the fly wheel. Unlike the "STOP" control, however, the "PPAF" control does not cut power to the trim press; it merely arrests the movement of the cutting mechanism.

On March 12, 1991, Grace Cummins, an employee of Mullinix, was operating the trim press on Mullinix's production line. She and another employee were setting up the press for the next production run. In the process, Ms. Cummins opened the scrap chute door and attempted to remove the scrap material that had accumulated. As she reached into the scrap chute, three of her fingers were severed by the trim press. Ms. Cummins later testified that, although she had intended to press the "STOP" control before opening the scrap chute door, she inadvertently had pressed the "PPAF" control instead. Following the accident, Mullinix's maintenance department inspected the trim press and discovered that the PPAF limit switch was out of adjustment. As a result, the adjustable arm of the limit switch was not in position to contact the cam on the rotating flywheel, and the brake did not activate.

Ms. Cummins brought this products liability action against Lyle, the manufacturer of the trim press. Before trial, Lyle filed a motion in limine to exclude certain testimony from one of the plaintiff's experts, Dr. Thomas Carpenter. Specifically, Dr. Carpenter was prepared to testify as follows concerning the feasibility of alternative designs for the braking system on the trim press: (1) the trim press ought to have been equipped with a "noncontact" limit switch instead of the mechanical limit switch actually installed; (2) a parallel time relay could have been wired to the PPAF circuit to shut down the trim press if the limit switch failed; and (3) the limit switch installed in the trim press ought to have had a longer useful cycle life. In addition, Dr. Carpenter was prepared to testify that the warnings and the instruction manual for the trim press were inadequate because they did not describe many of the controls for the press and because they did not explain the function of the trim press. The district court denied Lyle's motion to exclude these portions of Dr. Carpenter's testimony subject to reconsideration at trial.

On the second day of trial, the district court held a hearing to inquire into Dr. Carpenter's qualifications and the basis for his testimony. Following the hearing, the district court ruled that, although Dr. Carpenter would be permitted to testify with respect to certain other matters, he would not be permitted to offer an expert opinion concerning either the adequacy of warnings and the instruction manual or the feasibility of alternative designs for the trim press. The trial resumed and, during the direct examination of Dr. Carpenter, plaintiff's counsel began to question him on matters foreclosed by the district court's earlier order. When Lyle objected, Ms. Cummins presented a motion to reconsider the earlier order. The district court denied the motion but allowed Ms. Cummins to make a second offer of proof.

During the second offer of proof, Dr. Carpenter testified on direct examination that the cycle life of the limit switch used in the trim press "is indicated to be about three million cycles." Tr. at 91 (May 4, 1994). Dr. Carpenter opined that three million cycles is too short a cycle life for a machine designed to be subjected to a production schedule as rigorous as Mullinix's; the limit switch would last only a few months. On cross-examination, Dr. Carpenter testified that he had obtained the three million cycle figure in conversations with "several representatives from Allen Bradley," the manufacturer of the limit switch. Tr. at 5 (May 5, 1994). He admitted, however, that he could not identify the persons with whom he had spoken. Dr. Carpenter also admitted that he had not supplemented his discovery responses to include the three million cycle figure or otherwise disclosed this information to Lyle.

Applying Rule 702 and Daubert v. Merrell Dow Pharmaceuticals, Inc., 113 S.Ct. 2786 (1993) to the facts of this case, the district court took the view that Dr. Carpenter lacks a reliable basis for his proposed opinion. In evaluating the subject matter of Dr. Carpenter's testimony, the court found significant the fact that he had never tested his alternative designs and warnings or read any studies of such tests. The district court further noted that Dr. Carpenter does not have practical knowledge concerning the use of the alternative components in an industrial, machine-tool production environment.

Following the second offer of proof, the district court ruled that Dr. Carpenter would not be permitted to testify with respect to the cycle life of the limit switch because (1) Ms. Cummins had failed to disclose the cycle life of the switch as a basis for Dr. Carpenter's opinion prior to trial, see Fed.R.Civ.P. 26(a)(2)(B), and had failed to supplement Dr. Carpenter's discovery responses to reflect this newly-discovered information, see Fed.R.Civ.P. 26(e); and (2) the cycle life information amounts to unreliable hearsay which would not be reasonably relied upon by experts in the particular field, see Fed.R.Evid. 703. Ms. Cummins filed a "Renewed Motion to Reconsider," but the district court declined to modify its order. The jury returned a verdict in favor of Lyle. Ms. Cummins now appeals the rulings of the district court excluding Dr. Carpenter's proposed testimony.

II. [12] DISCUSSION A.

Ms. Cummins first submits that the district court improperly excluded Dr. Carpenter's testimony concerning the feasibility of alternative design and the adequacy of warnings and instructions for the trim press. The district court was of the view that Dr. Carpenter, whose education and experience lie in the area of agricultural engineering, is not qualified to give expert testimony concerning the adequacy of instructions and warnings and concerning the feasibility of alternative designs in the context of high-speed factory equipment used for mass production. The court further reasoned that Dr. Carpenter lacks a reliable basis upon which to render an opinion concerning alternative designs.

We need not decide whether the district court abused its discretion in refusing to qualify Dr. Carpenter as an expert with respect to his proposed testimony. Even if we were to assume that Dr. Carpenter was qualified to render an expert opinion on the feasibility of alternative designs and warnings for the trim press, we still would have to conclude that the district court acted within its discretion when it decided to exclude his testimony. The independent requirement that the subject matter of his expert testimony pertain to "scientific, technical, or other specialized knowledge" remains, Fed.R.Evid. 702, and the district court was justified in concluding that Dr. Carpenter's testimony, predicated on his own untested observations, was not admissible.

In reviewing the district court's application of Rule 702, our inquiry is governed by Daubert v. Merrell Dow Pharmaceuticals, Inc., 113 S.Ct. 2786 (1993). We first undertake a de novo review of whether the district court properly followed the analytical framework established in Daubert. Bradley v. Brown, 42 F.3d 434 (7th Cir. 1994); see Mars v. United States, 25 F.3d 1383, 1383-84 (7th Cir. 1994). "Provided the district court adhered to Daubert's parameters, we will not disturb the district court's findings unless they are manifestly erroneous." Bradley, 42 F.3d at 437. The decision to allow or to exclude expert testimony is committed to the sound discretion of the district court. See id. at 436-37.

Our cases interpreting Daubert have endorsed a two-step analysis for the district courts to use in evaluating expert testimony under Rule 702. First, the district court must determine whether the expert's testimony is reliable. As the court said in Bammerlin v. Navistar International Transportation Corp., 30 F.3d 898 (7th Cir. 1994), a "district judge should assure himself, before admitting expert testimony, that the expert knows whereof he speaks." Id. at 901. In the context of theoretical and applied science, this requirement places on the court the obligation to ensure that the proffered testimony pertains to scientific knowledge. Deimer, 58 F.3d at 344; Porter v. Whitehall Labs., 9 F.3d 607, 614 (7th Cir. 1993). "This task requires the district court to consider whether the testimony has been subjected to the scientific method; it must rule out `subjective belief or unsupported speculation.'" Deimer, 58 F.3d at 344 (quoting Porter, 9 F.3d at 614 (quoting in turn Daubert, 113 S.Ct. at 2795)). Second, the district court has to determine "whether evidence or testimony assists the trier of fact in understanding the evidence or in determining a fact in issue." Porter, 9 F.3d at 616; see Deimer, 58 F.3d at 345.

We begin with the first of the two inquiries mandated by Daubert — that the proposed testimony pertain to "scientific knowledge." The Supreme Court in Daubert, noted that Rule 702 "clearly contemplates some degree of regulation of the subjects and theories about which an expert may testify." Daubert, 113 S.Ct. at 2795. In carrying out this mandate, the Supreme Court held, the district court must ensure that "any and all scientific testimony or evidence admitted is not only relevant, but reliable." Daubert, 113 S.Ct. at 2795. An expert scientific opinion must be grounded in the "methods and procedures of science," and must consist of more than simply "subjective belief or unsupported speculation." Id. The Court explained:

[A]n inference or assertion must be derived by the scientific method. Proposed testimony must be supported by appropriate validation — i.e., "good grounds," based on what is known. In short, the requirement that an expert's testimony pertain to "scientific knowledge" establishes a standard of evidentiary reliability.

Id. As another panel of this court put it, "a district court asked to admit scientific evidence must determine whether the evidence is genuinely scientific, as distinct from being unscientific speculation offered by a genuine scientist." Rosen v. Ciba-Geigy Corp., 78 F.3d 316, 318 (7th Cir. 1996).

As our cases applying Daubert have recognized, the Supreme Court has articulated several nonexclusive guide-posts to assist the district courts in determining whether expert testimony fairly can be characterized as a scientific opinion: (1) whether the proffered conclusion lends itself to verification by the scientific method through testing; (2) whether it has been subjected to peer review; (3) whether it has been evaluated in light of the potential rate of error of the scientific technique; and (4) whether it is consistent with the generally accepted method for gathering the relevant scientific evidence. Deimer, 58 F.3d at 344; see Gruca v. Alpha Therapeutic Corp., 51 F.3d 638, 643 (7th Cir. 1995); Porter, 9 F.3d at 615. The inquiry is a flexible one; the focus must be solely on principles and methodology, not on the conclusions they generate. Daubert, 113 S.Ct. at 2797.

The first and most significant Daubert factor is whether the proffered opinion has been subjected to the scientific method. Bradley, 42 F.3d at 438. Ms. Cummins submits that the district court erred in requiring Dr. Carpenter to have tested his proposed alternative designs because his designs involve only simple components and widely-accepted engineering principles. During the offer of proof, Dr. Carpenter testified that the non-contact limit switch and the time-delay relay that he proposes are "off-the-shelf" components for "these types of applications." Tr. at 60 (Apr. 26, 1994). He also stated that other circuits in the trim press already incorporate the time-delay relay that he proposes for the PPAF circuit. Under these circumstances, Ms. Cummins asserts, no testing or further assessment of the alternative designs was necessary.

Our cases have recognized the importance of testing in alternative design cases. In Deimer v. Cincinnati Sub-Zero Prods., Inc., 58 F.3d 341 (7th Cir. 1995), the plaintiff, a surgical nurse, was injured when she stumbled on the power cord of a heavy medical device, causing the machine to fall on top of her. The district court granted the defendant's motion to strike the testimony of the plaintiff's expert, Dr. Roland Ruhl, on the issue of whether the power cord wrap had been defectively designed. The district court in Deimer took the view that Dr. Ruhl had failed to substantiate his opinion on the basis of any scientific research. On appeal, we upheld this exercise of the district court's discretion. We noted that Dr. Ruhl "had proffered unverified statements that were unsupported by any scientific method." Deimer, 58 F.3d at 345. This type of unsubstantiated testimony, we explained, "provides no basis for relaxing the usual first-hand knowledge requirement of the Federal Rules of Evidence on the ground that the expert's opinion has a reliable basis in knowledge and experience of his discipline." Id. (citing Daubert, 113 S.Ct. at 2796); see Rosen, 78 F.3d at 319 (upholding the exclusion of "bottom line" scientific testimony unsupported by experimental, statistical, or other scientific data).

Similarly, in O'Conner v. Commonwealth Edison Co., 13 F.3d 1090 (7th Cir.), cert. denied, 114 S.Ct. 2711 (1994), we found no error in the district court's determination that the testimony of the plaintiff's expert witness, a physician, was inadmissible. The physician, who had relied on his personal observation to determine that the plaintiff's cataracts were induced by radiation, produced no "personal study or experiment that otherwise would justify his conclusion that [the plaintiff's] cataracts are radiation-induced." Id. at 1107. Because the witness thus failed to establish a proper scientific basis for his opinion, we held, the district court's exclusion of his testimony was not an abuse of discretion. Id.

We do not mean to suggest, of course, that hands-on testing is an absolute prerequisite to the admission of expert testimony. Rule 702 is designed to ensure that, when expert witnesses testify in court, they adhere to the same standards of intellectual rigor that are demanded in their professional work. See Rosen, 78 F.3d at 318; Bammerlin, 30 F.3d at 901. This objective can be accomplished in a number of different ways, including through the review of experimental, statistical, or other scientific data generated by others in the field. Indeed, in Porter v. Whitehall Labs., Inc., 9 F.3d 607 (7th Cir. 1993), we acknowledged that "there may be a situation in which personal experiments or observations meet the requirements of Daubert." Id. at 614 n. 6. As in Porter, however, the opinions offered by Dr. Carpenter in this case clearly lend themselves to testing and substantiation by the scientific method. The district court clearly acted well within its discretion in concluding that the absence of such testing indicated that the witness' proffered opinions could not fairly be characterized as scientific knowledge.

It is important to note that Dr. Carpenter's proposed testimony was not limited to a discussion of the existence of the "off the shelf" components and their operational characteristics. Dr. Carpenter was prepared to render the much broader opinion that Lyle ought to have incorporated the alternative switch designs and warnings into the Lyle trim press. There are a number of considerations which must inform such a conclusion. These include, but are not limited to, the degree to which the alternative design is compatible with existing systems and circuits; the relative efficiency of the two designs; the short- and long-term maintenance costs associated with the alternative design; the ability of the purchaser to service and to maintain the alternative design; the relative cost of installing the two designs; and the effect, if any, that the alternative design would have on the price of the machine. Many of these considerations are product and manufacturer-specific, and most cannot be determined reliably without testing.

Our review of the record in this case supports the district court's conclusion that Dr. Carpenter did not conduct any testing to substantiate his opinion that the alternative test designs feasibly could have been incorporated into the Lyle trim press. During the first offer of proof, Dr. Carpenter agreed that testing is a necessary part of the design process. He acknowledged, however, that he had not tested any of the machinery at issue in this case:

Q. You agree, do you not that the design process of a piece of machinery is more complicated than making instantaneous judgments; there has to actually be testing of that design process, is that correct?

A. Absolutely.

Q. And designing hazards of the machinery requires that kind of testing, doesn't it?

A. That's correct.

Q. You haven't done any testing of the designs in relationship to your work in this case, have you?

A. In general, we have not been talking about designs per se. We've been talking about some components that are involved.

Q. Have you done any testing of machinery or machinery components in relationship to your work in this case?

A. No.

Tr. at 73 (Apr. 26, 1994). Dr. Carpenter also acknowledged that he had not read any studies, surveys or analyses of how this type of machinery is designed, manufactured or used. With respect to the adequacy of warnings, Dr Carpenter testified that he had read texts that discussed warnings, but that he did not know what warnings would have been effective in the absence of testing.

Nor is there any indication in the record that the basis for Dr. Carpenter's opinion satisfies the other guideposts identified by the Supreme Court in Daubert. The second of these touchstones is whether the theory has been subjected to peer review. By his own admission, Dr. Carpenter has not published any writing or study concerning the incorporation of any of the alternative designs into a trim press such as Lyle's. Given the nature of Dr. Carpenter's opinion, moreover, the district court had no occasion to address the third guidepost articulated in Daubert: whether the expert testimony has been evaluated in light of the potential rate of error of the scientific technique. Certainly, as we already have noted, the testing methodology that the district court found lacking is consistent with the generally accepted method for gathering and evaluating data in alternative design cases. See Deimer, 58 F.3d at 344; see also United States v. Pierre, 49 F.3d 241, 243 (7th Cir. 1995). Under these circumstances, we are unable to conclude that the district court abused its discretion when it ruled that Dr. Carpenter's conclusions were not "derived from the scientific method." Daubert, 113 S.Ct. at 2795; Porter, 9 F.3d at 613-14.

The Supreme Court's second directive in Daubert is that, when confronted with the proffer of expert testimony, a district court must determine whether the evidence or testimony assists the trier of fact in understanding the evidence or in determining a fact in issue. Stated differently, the scientific testimony must "fit" the issue to which the expert is testifying. Daubert, 113 S.Ct. at 2796; Deimer, 58 F.3d at 345; Bradley, 42 F.3d at 437. Because the district court determined that Dr. Carpenter could not render a scientific opinion with respect to the two matters at issue, we need not address extensively this latter directive of Daubert.

We must conclude, therefore, that, with respect to the proffered testimony as to the feasibility of alternate designs and the adequacy of warnings and instructions for the trim press, the district court properly adhered to the methodology established in Daubert. Our examination of the record confirms that the district court carefully performed its gatekeeping function under Rule 702 and heeded the Supreme Court's admonition that "the focus . . . must be solely on the principles and methodology, not on the conclusions they generate." Daubert, 113 S.Ct. at 2797. Noting that it was exercising "an abundance of caution and latitude," Tr. at 60 (May 4, 1994), the court permitted the plaintiff to make two separate offers of proof, totaling some twelve to fifteen hours of testimony. During each offer of proof, Ms. Cummins was given ample opportunity to establish Dr. Carpenter's qualifications and to present the substance of his testimony. Before ruling on Lyle's objections to Dr. Carpenter's testimony, the district court reviewed Federal Rule of Evidence 702 and Daubert. The court specifically acknowledged its "gatekeeping role" in assessing the reliability of the basis for Dr. Carpenter's opinion. See Daubert, 113 S.Ct. at 2798; id. at 2800 (Rehnquist, C.J., concurring in part and dissenting in part) ("I do not doubt that Rule 702 confides to the judge some gatekeeping responsibility in deciding questions of the admissibility of expert testimony."); Bradley, 42 F.3d at 437-38. The court examined Dr. Carpenter's qualifications in the area of industrial engineering, the possibility of applying the scientific method to his alternative designs and warnings, and the presence or absence of peer review for his proposal. Finally, in making its ruling, the district court was careful to delineate the areas with respect to which Dr. Carpenter would be permitted to testify and those on which he would not.

We hold, therefore, that the district court did not abuse its discretion in refusing to allow Dr. Carpenter to render an expert opinion regarding the adequacy of warnings and the instruction manual and the feasibility of alternative designs for the trim press.

B.

Ms. Cummins also submits that the district court erred in excluding evidence concerning the useful life of the limit switch. The district court excluded this testimony on two grounds. First, the plaintiff failed to disclose the cycle life of the switch as a basis for Dr. Carpenter's opinion prior to trial, see Fed.R.Civ.P. 26(a)(2)(B), and failed to supplement Dr. Carpenter's discovery responses to reflect this newly discovered information, see Fed.R.Civ.P. 26(e). Second, the court found, the source of Dr. Carpenter's information amounts to unreliable hearsay that would not be reasonably relied upon by experts. See Fed.R.Evid. 703. We review the exclusion of evidence on each of these grounds for an abuse of discretion. See Fed.R.Civ.P. 26 advisory committee's note; see, e.g., Patel v. Gayes, 984 F.2d 214, 221 (7th Cir. 1993) ("Under the Federal Rules, the district court has the discretion to impose sanctions on a party if that party fails to meet the requirements of Rule 26."); Gong v. Hirsch, 913 F.2d 1269, 1272-73 n. 3 (7th Cir. 1990) (reviewing the district court's exclusion of expert testimony under Rule 703 for an abuse of discretion).

We are unable to conclude that the district court abused its discretion when it excluded Dr. Carpenter's proposed testimony on the cycle life of the limit switch. The plaintiff failed to disclose, in violation of two separate discovery provisions, the fact that Dr. Carpenter's opinion would rely on a cycle life figure of three million cycles. See Walsh v. McCain Foods Ltd., 81 F.3d 722, 727 (7th Cir. 1996). Allowing Dr. Carpenter to render an expert opinion based on newly acquired evidence would have imposed an unfair burden on Lyle to locate and depose a witness who is competent to give testimony on the cycle life of the limit switch. Even if Lyle were granted a continuance to locate and depose a witness familiar with the cycle life of the limit switch, the prejudice to Lyle from the admission of Dr. Carpenter's evidence would not have been fully cured. The prejudice experienced by Lyle from the delay in disclosure would have been compounded by the fact that, even at the time that the discovery sanction was imposed, Dr. Carpenter was unable to identify the sources of his information so that they could be questioned by Lyle. This failure clearly would have prejudiced Lyle's ability to cross-examine Dr. Carpenter. Under these circumstances, the district court reasonably could have concluded that the prejudice accruing to Lyle from the plaintiff's failure to disclose this information warranted exclusion of the cycle life testimony.

The district court's ruling also finds ample support in Federal Rule of Evidence 703. Under Rule 703, expert opinion evidence is not objectionable even though the expert has relied upon underlying information which is itself inadmissible, as long as the inadmissible information is of a kind "reasonably relied upon by experts in the field." Fed.R.Evid. 703. In the present case, the only evidence in the record relating to the reliability of cycle life information came from Gerald Schwartz, the president of Lyle. Schwartz testified that cycle life information should be obtained directly from the manufacturer of the limit switch.

Dr. Carpenter testified that he obtained the three million cycle figure in conversations with "several representatives from Allen Bradley," the manufacturer of the limit switch. Tr. at 5 (May 5, 1994). He admitted, however, that he could not recall the names of the individuals that he had spoken with or the dates on which he had obtained this information.

The district court was entitled to conclude that there has been an inadequate showing that the hearsay statements relied upon by Dr. Carpenter in formulating his opinion are "of a type reasonably relied upon by experts in the field." Fed.R.Evid. 703. Dr. Carpenter obtained the cycle life information from unidentified individuals on unspecified dates. There is no indication in the record that experts in Dr. Carpenter's field reasonably rely on statements of this kind. Accordingly, we are unable to conclude that the district court abused its discretion when it prevented Dr. Carpenter from rendering an opinion based on the three million cycle figure.

Conclusion

For the foregoing reasons, the judgment of the district court is affirmed.

AFFIRMED