Del Mar Engineering Laboratories (Del Mar) is the assignee of patent # 3,267,934. Claims one through eight of this patent relate to a device for measuring electrically the “ST segment” of the human heartbeat. Physio-Tronics, Inc., is the southwest distributor of Quinton Instruments, a Washington corporation. Del Mar brought this suit against Physio-Tronics alleging infringement of its patent by two devices produced by Quinton Instruments.

Physio-Tronics contends that the patent is invalid: first, because William Thornton, the original patentee, deceptively failed to disclose that he had made public use of a ratemeter, an unrelated device covered by claims 15 and 16 of the same patent; second, because the patent claims relating to the “ST segment” measuring device fail under the statutory defenses of anticipation and obviousness; finally, it contends that if the patent is valid, there was no infringement by the Quinton equipment. The trial judge concluded that the patent was valid and that it had been infringed. We affirm.

I. INVALIDITY OF THE PATENT FOR FAILURE TO DISCLOSE A PUBLIC USE

The statutory bar of 35 U.S.C. § 102 (b) prevents entitlement to a patent if there is a public use or sale “more than one year prior to the date of the application for patent .. . . ”

Thornton filed his patent application on September 20, 1962. Physio-Tronics points to Thornton’s exhibition of a ratemeter at two meetings of the American Medical Association, its use on patients, and an alleged sale of the device to Duke Medical School as a sale and public uses that occurred before the date barred by statute, September 20, 1961. Physio-Tronics contends that the entire Del Mar patent, including the patent claims to the “ST segment” measuring device, is invalid because Thornton committed fraud by failing to disclose the ratemeter uses to the patent office.

In 1976, prior to the time the case went to trial, Del Mar filed a disclaimer to the ratemeter claims 15 and 16, but Physio-Tronics contends that the disclaimer cannot save the patent. In its view, the patent was originally filed with “deceptive intention” within the meaning of 35 U.S.C. § 253 , so any later disclaimer cannot save the patent’s initial invalidity.

Physio-Tronics’ argument can be accepted only if the public uses were in fact barred by statute, and if it can be shown that Thornton’s failure to inform the patent office was “culpable” or in “bad faith.” See Carpet Seaming Tape Licensing v. Best Seam, Inc., 616 F.2d 1133 , 1138 (9th Cir. 1980) (a showing of culpability is required to establish fraud on the Patent Office).

Turning first to the question of public use, it should be noted that the statute does not bar genuine experimental uses. The policy underlying the statutory bar of § 102(b) is to prevent an inventor from extending the period of a patent monopoly by commercially exploiting the invention before applying for a patent. See, e. g., Pickering v. Holman, 459 F.2d 403, 406 (9th Cir. 1972).

When an invention has been reduced to practice, further public testing and demonstration may well support an inference that the inventor’s intent is to exploit the invention, but “there may be an experimental use even following reduction to practice where the experiments are an attempt to further refine the device.” Cataphote Corp. v. DeSoto Chemical Coatings, Inc., 235 F.Supp. 931, 934 (N.D.Cal.1964). See Elizabeth v. Pavement Co., 97 U.S. 126, 135-36 , 24 L.Ed. 1000 (1877). An inventor may also sell an unpatented device and not be barred by statute. The Ninth Circuit rule in this regard is that a sale is permitted if there is “an express or clearly implied condition that the sale or offering is made primarily for experimental use.” Robbins Co. v. Lawrence Manufacturing Co., 482 F.2d 426, 433 (9th Cir. 1973).

Whether a public use is for the purpose of exploiting an invention or whether it is a bona fide experimental use turns largely on the intent or motivation of the inventor. Elizabeth v. Pavement Co., 97 U.S. 126, 135 , 24 L.Ed. 1000 (1877); Pickering v. Holman, 459 F.2d at 406 . As such, it is a question of fact to be determined by the trial court and we will reverse the trial judge’s determination only if it is clearly erroneous. See Micro-magnetic Industries, Inc. v. Advance Auto Sales Co., 488 F.2d 771, 773 (9th Cir. 1973).

Thornton developed the ratemeter during the period between 1959 and 1961. As a prototype, Thornton tested it frequently at the hospital of the University of North Carolina. The ratemeter was not in routine use, but was utilized only in clinical re-* search at the hospital. Thornton testified * that he wanted to assess the practical utility of the device and to look for shortcomings. Although he continued to make modifications to the ratemeter, he testified that it generally performed well. Additionally, the device was apparently used at Duke Medical School at about this time, again on a select group of patients.

Thornton also submitted a resume describing the ratemeter to the American Medical Association. It was selected as one of ten or twelve other exhibits to be shown at a student meeting of the American Medical Association held in Chicago in either late 1960 or early 1961. Thornton’s invention, which won first prize, was exhibited along with the other student scientific exhibits in an area separated from the commercial exhibits. Thornton was subsequently invited to show the ratemeter again at a general meeting of the American Medical Association in New York in June 1961. On both occasions the American Medical Association provided free space to show the exhibit.

The trial judge found that all the uses that took place prior to the date of the statutory bar occurred while the ratemeter was in a developmental stage. He concluded that the exhibition of the ratemeter as a student experiment was not inconsistent with an experimental use because Thornton did not intend to exploit it commercially by so exhibiting it. These findings are supported by the evidence and are not clearly erroneous.

As to the sale to Duke University that allegedly occurred prior to September 1961, Thornton could only give what he described as a “very wide estimate” of when it might have taken place. He testified that it was “possibly” as early as the summer of 1961. Although Thornton’s vague testimony might have been the basis for a finding that the sale occurred prior to September 1961, in the absence of other testimony the trial judge’s conclusion that the device was not offered for sale until January 1962, when it first became commercially available from Del Mar, is not clearly erroneous. It is, therefore, unnecessary to consider whether there was an implied condition in the sale of the ratemeter that it be used on an experimental basis. Because there was no public use or sale within the meaning of 35 U.S.C. § 102 (b), it is also unnecessary to consider whether Thornton deceptively failed to inform the patent office.

The patent is not invalid on the ground that Thornton failed to disclose that the ratemeter claims were in public use.

II. INVALIDITY FOR ANTICIPATION AND OBVIOUSNESS

Physio-Tronies next raises the two statutory defenses of anticipation, 35 U.S.C. § 102 (a), and obviousness, 35 U.S.C. § 103 . Before considering these defenses, it is first necessary to understand the claims of the Thornton patent that are at issue in this case, and the prior art.

A. Thornton’s ST Segment Measuring System

The heart’s nervous system produces electrical voltages which vary in amplitude during the course of a heartbeat. A plot of voltage amplitude over time yields a wave form known as an electrocardiogram or EKG. An idealized EKG begins with a shallow undulation known as the “P” wave, followed by a flat line of no voltage change known as the “PQ interval.” There is then a rapid rise and fall in voltage known as the “R” wave, which is again followed by a flat segment of no voltage change known as the “ST segment.” Finally at the end of a heartbeat there is another shallow undulation known as the “T” wave. The following diagram illustrates a typical EKG for a heartbeat:

Prior to Thornton’s work, it was known that depression of voltage during the ST segment of the cardiogram could be an indication of heart disease. A cardiologist using paper recording tape and a ruler could draw a line along the bottom of the EKG waveform. From this reference line, it was possible to measure the height of the ST segment. This technique was useful for a small number of heartbeats, but as a practical matter it was nearly impossible to perform for a large number. The paper tape necessary for a full day of monitoring, for example, might be more than a mile long. Furthermore, cardiologists apparently consider it important to measure the ST segment under a variety of conditions of heart stress, so that a large number of samples is desirable.

Between 1960 and 1961, Thornton, who was then a medical student, attempted to solve the problem of measuring the ST segment continuously for a large number of heartbeats by using electronic circuitry. The essence of his technique was to measure the voltage during the first flat portion of the EKG, the PQ interval, and use this as a reference to measure the ST segment. Electrically this would work much like the cardiologist drawing a baseline underneath the EKG.

The easiest method of comparing the PQ interval with the ST segment would be for the circuit to somehow “recognize” the increase in voltage caused by the P wave and to store the voltage that occurred several milliseconds after the P wave during the flat portion of the PQ interval. This sam pie could then be compared with the ST segment. The P wave, however, is a relatively small increase in voltage which may cause it to be confused with background noise. Also when the heart beats rapidly, the beginning P wave of one beat may overlap the T wave of the last preceding beat. Thornton found that these features of the P wave made it a poor indicator. The only unambiguous reference point in the cardiogram is the spike in voltage caused by the R wave.

As Thornton testified, the problem with using the R wave is that electrical voltages can only be measured as they occur. When the R wave occurs during the heart’s cycle, the PQ interval for that beat has already passed, so it is not possible to go back in time to measure it. Del Mar claims that the unique feature of the Thornton patent is that it effectively goes back in time to measure the PQ interval once the circuit recognizes the R wave.

Although somewhat of an over-simplification, the circuit does this by taking samples continuously at discrete intervals. The circuit stores a sample for a number of milliseconds, and then releases it. There is no synchronization between the time when samples are taken and heartbeats, so samples are effectively taken at random. Many samples will thus be taken at other times than the PQ interval and will not be of use, but some samples will be taken during the PQ interval. When the R wave does occur, the circuit can then call up a sample that is being stored. By timing the interval between each sample, Thornton was able to insure that a sample would be available that had been taken at some time during the PQ interval. The ST segment can be recognized because it follows immediately after the R wave. By comparing the sample from the PQ interval with the following ST segment, Thornton could determine the height of the ST segment.

B. The Prior Art

Physio-Tronics claims that the most relevant prior art was the work of Dr. Caseres and the Moore patent. The Caseres project, which was developed under the auspices of the Public Health Service, involved the use of standard digital computers to measure the EKG. The system sampled an EKG at the rate of 625 per second. It then converted the samples from voltage magnitudes to numbers by a device known as an analog to digital converter. Number values were then stored in the memory bank of a standard computer. Caseres programmed the computer to extract various measurements from the EKG, including the height of the ST segment.

The trial judge found a number of differences between the Thornton patent and the Caseres system. Thornton used a few circuits to sample and store voltages of various magnitude rather than digital values. He maximized the efficiency of the storage capacity he had by sampling continuously at timed intervals and releasing unused samples. His system did not involve the use of standard computers. The Caseres system, at the time it was developed, required manual preparation of data. It took 64 minutes to completely process the results of a single heartbeat. Thornton’s device produced results instantaneously, but it did not have the capability to measure other characteristics of the EKG which the Caseres system did. Thornton produced a device that was portable, relatively inexpensive, and that could be used on an ambulatory patient. Caseres found, as he stated in his deposition, that to attempt to monitor patients directly with his system “was impractical from a logistical viewpoint.” The computers required air conditioning and were so big “you couldn’t put the patient in the same room .... ”

The Moore patent involved a circuit that sampled and held a voltage, somewhat similar to Thornton’s device. The Moore patent, however, involved the use of a television set circuit, but more importantly, it did not involve the problem of storing a sample before a triggering event, like the Thornton device, because it used an event which had already occurred to achieve timing.

C. Anticipation

35 U.S.C. § 102 provides that:

A person shall be entitled to a patent unless—

To overcome the defense of anticipation, it is only necessary for the patentee to show some tangible difference between the invention and the prior art. We have described anticipation several times as

Walker v. General Motors Corp., 362 F.2d 56, 58 (9th Cir. 1966) quoted in Saf-Gard Products, Inc. v. Service Parts, Inc., 532 F.2d 1266, 1270 (9th Cir.), cert. denied, 429 U.S. 896 , 97 S.Ct. 258 , 50 L.Ed.2d 179 (1976).

From the description above, and as the trial judge found, it seems obvious that the physical differences in the operation of the Caseres system and Moore patent would bar an anticipation defense. Physio-Tronics does not dispute these facts. Its argument is that if one compares only the broad language of the Thornton patent claims with the description of the Caseres system the Thornton device seems identical.

The Ninth Circuit rule, however, is that

Illinois Tool Works, Inc. v. Brunsing, 389 F.2d 38, 40 (9th Cir. 1968). This language, as well as similar language in United States v. Adams, 383 U.S. 39, 49 , 86 S.Ct. 708, 713 , 15 L.Ed.2d 572 (1966), indicates that it is permissible to examine the specifications to understand properly the scope of a patent claim.

Physio-Tronics, while apparently recognizing this rule, contends that in this case the court cannot look beyond the language of the patent claims. In its answer to Del Mar’s complaint, Physio-Tronics alleged, as an affirmative defense, that the patent claims were too uncertain and ambiguous to support an act of infringement. At trial the judge dismissed this defense because there was no evidence to support it. Physio-Tronics now reasons that if the trial judge thought the claim language was clear enough to support an act of infringement, it is not necessary to look beyond the claims to evaluate a defense of anticipation.

This argument is without merit. First, Physio-Tronics reads far too much into the trial judge’s dismissal of its affirmative defense. The trial judge may well have looked to the specifications in deciding that the claim language was clear enough to support an act of infringement. There is nothing in the record that would indicate that he did not. Second, Physio-Tronics has not cited any authority which would support its narrow view of when it is proper to examine the specifications; we have found none, and reject the argument. We conclude, as the trial judge did, that the Thornton patent was not anticipated by the prior art.

D. Obviousness

In addition to the requirement that to be patentable an invention must tangibly differ from prior art, it is also required that the differences are not obvious. This requirement is contained in 35 U.S.C. § 103 , which prevents the granting of a patent monopoly if

Ultimately, the question of obviousness or nonobviousness is a legal conclusion, but the Supreme Court has considered the nature of the factual inquiry which should precede such a finding.

Graham v. John Deere Co., 383 U.S. 1, 17 , 86 S.Ct. 684, 693 , 15 L.Ed.2d 545 (1966).

The scope and content of the prior art Physio-Tronics cites as most relevant, and the differences between this prior art and the Thornton patent, have been discussed. As to the level of skill exercised in the field of electronics at the time Thornton developed the ST computer, Physio-Tronics relies principally upon the testimony of its expert witness, Dr. Farmer. Farmer considered the Thornton device to be obvious in light of the Caseres work and the Moore patent. The trial court, however, placed little reliance upon his testimony. Farmer seemed to consider the Thornton device mainly in light of the language of the patent claims without looking at the specifications. He conceded that looking beyond the claims “was another matter” and acknowledged differences between the Caseres system, the Moore patent and Thornton’s device.

Farmer’s testimony was also contradicted by Sanctuary, who as Del Mar’s Chief Engineer during the early 1960’s was familiar with the level of skill in the electronics industry. He testified that in his opinion the Thornton technique for effectively going back in time from the R wave to obtain a sample from the PQ interval was “an extremely clever idea” and not a readily apparent approach.

Issued patents are entitled to a statutory presumption of validity. 35 U.S.C. § 282 . This presumption can be rebutted only by “clear and convincing evidence,” Saf-Gard Products, Inc. v. Service Parts, Inc., 532 F.2d 1266, 1271 (9th Cir.), cert. denied, 429 U.S. 896 , 97 S.Ct. 258 , 50 L.Ed.2d 179 (1976), but it does not apply where a defendant has shown that the patent office was not informed of the most relevant prior art. See Houston v. Polymer, 637 F.2d 617, 619 (9th Cir. 1980); Cool-Fin Electronics Corp. v. International Electronic Research Corp., 491 F.2d 660 , 661 n.2 (9th Cir. 1974). Physio-Tronics argues that the presumption of validity does not apply here because Thornton failed to advise the patent office of the Moore patent and the Caseres work. Arguably these two prior art references were more relevant than the other work cited in the patent application, which consisted mainly of electronic medical devices. However, the trial judge concluded that the two references cited by Physio-Tronics were not relevant and could not overcome the presumption of validity.

We need not decide whether the presumption of validity applies. The subsidiary findings of fact made by the trial judge relating to the prior art, the differences in the prior art, and the level of skill in electronics are all supported by the evidence. The evidence adequately demonstrates, without the aid of a presumption, that the Thornton ST measurement technique was not obvious. We also find no merit in Physio-Tronics’ contention that Thornton committed fraud on the patent office by failing to inform them of the Caseres system and Moore patent.

III. INFRINGEMENT

Physio-Tronics finally contends that if one reads the patent claims in light of the specifications in order to avoid invalidating the patent, the claims are so narrowed that they cannot be read to include the Quinton equipment. We find no merit to this argument. The essential feature of the Thornton device, considered by Sanctuary to be its most unique feature, is its technique for measuring the relevant PQ interval before a triggering event. In its brief, PhysioTronics concedes that “the ST level measurement circuits of the two devices [i. e., the Quinton and Thornton devices] were the same.” The chief difference between the Thornton device and the Quinton device is that the Quinton device averaged the ST segment measurement over 20 heartbeats rather than providing a readout for each heartbeat. The circuitry otherwise performs in substantially the same manner.

Infringement requires the existence of a “substantial identity of function, means, and result.” Pursche v. Atlas Scraper & Engineering Co., 300 F.2d 467, 482 (9th Cir. 1962). The trial judge found that the specifications and claims of the Del Mar patent could be read to include the Quinton models and concluded that the Del Mar claims had been infringed. This conclusion is supported by the evidence. Accordingly, the judgment of the trial court is

AFFIRMED.

. 35 U.S.C. § 253 provides in part: Whenever, without any deceptive intention, a claim of a patent is invalid the remaining claims shall not thereby be rendered invalid. . Judge Crary, who tried the case in the district court, died before he rendered his decision. The parties then agreed to submit the case on the transcripts and other documents to Judge Curtis for decision. In addition to reviewing the record, Judge Curtis conducted a hearing to allow counsel for the parties to present their positions. Physio-Tronics now contends that because Judge Curtis did not try the case him- . self, but only relied upon the trial court record of Judge Crary, his findings of fact should not, on appeal, be accorded the degree of deference of the clearly erroneous standard. We considered and rejected this same argument in United States v. Mountain States Const. Co., 588 F.2d 259 , 264 n.5 (9th Cir. 1978), which we shall continue to follow here. Judge Curtis’ factual findings will be evaluated under the clearly erroneous standard.