Opinion
KENNARD, J.At petitioner’s 1997 trial for the murder of his wife, the evidence against him was circumstantial. One of many pieces of evidence linking petitioner to the crime was a postmortem photograph of the victim’s hand, depicting an indistinct, crescent-shaped lesion. The prosecution’s dental expert stated his opinion that the lesion was a human bite mark. And after comparing the bite mark to the distinctive arrangement of petitioner’s lower teeth, the dental expert stated that petitioner’s unusual dentition was consistent with the shape of the mark depicted in the photograph, and that he could not exclude petitioner’s teeth as a possible source of the mark. According to the expert, petitioner’s unusual dentition occurred in only 2 percent or less of the general population. The jury found petitioner guilty as charged. The Court of Appeal upheld the conviction.
In 2007, petitioner sought habeas corpus relief in the San Bernardino County Superior Court, claiming that new evidence established his innocence, and that his murder conviction was based on false evidence given at trial by the prosecution’s dental expert. -In a declaration supporting the petition, that expert stated that his trial testimony regarding the statistical frequency of petitioner’s dentition was not based on scientific data. Moreover, after examining photographs depicting other lesions on the victim’s body, the dental expert said he was no longer certain that the lesion on the victim’s hand was a bite mark.
*952Supporting declarations by other dental experts agreed, based on newly available computer technology, that the prosecution’s expert had testified inaccurately at trial.
The superior court issued an order to show cause. After an evidentiary hearing, the court granted habeas corpus relief; in the court’s view, petitioner had presented new evidence pointing unerringly to his innocence. The Court of Appeal disagreed. We granted petitioner’s request for review.
The most significant issue here is whether a conviction is based on “false evidence” (Pen. Code, § 1473, subd. (b)) when it depends in part on the opinion of an expert witness, and posttrial advances in technology have raised doubts about the expert’s trial testimony without conclusively proving that testimony to be untrue. We conclude that in such circumstances the expert’s trial testimony has not been shown to be “false evidence,” but that the information garnered from the technological advances may be presented as newly discovered evidence in support of habeas corpus relief. Habeas corpus relief should be granted only if the new evidence “ ‘point[s] unerringly to innocence or reduced culpability’ ” {In re Clark (1993) 5 Cal.4th 750, 766 [21 Cal.Rptr.2d 509, 855 P.2d 729]), a showing that petitioner here has not made.
I
A. Murder of Pamela Richards
Petitioner and his wife, Pamela, lived in a camper parked on their property in a remote area of San Bernardino County. They used a generator, kept in a small shed, for electricity. To access their home, one had to ascend a steep sand and gravel driveway. The couple kept several dogs on the property to ward off uninvited intruders; they also had several guns, which Pamela knew how to use.
At approximately 11:55 p.m. on August 10, 1993, Eugene Price telephoned the couple’s camper in response to a message Pamela had left on his answering machine around 7:00 or 7:30 that evening. Price had a sexual relationship with Pamela, who was planning to move with Price to an apartment in Ventura County. Petitioner answered the telephone call. He sounded stressed and agitated. When Price asked for Pamela, petitioner said she was dead. Price told petitioner to call 911. At 11:58 p.m., petitioner did so.
*953Because of the remote location of the property, San Bernardino Sheriff’s Deputy Mark Nourse did not arrive until shortly after 12:30 a.m. He testified that the property was “pitch black,” there were no lights, and there was no moonlight. Petitioner, who was dressed in blue jeans and a blue work shirt, gave this story: He had left work at 11:00 p.m. and arrived at the camper just before midnight. The generator was off when he arrived. The property was dark. The battery in the camper had lost its charge. He did not turn on the generator, and he had no light. He found his wife lying on the ground outside the camper. It was hard for him to see her body in the dark, but he realized she was dead when he rolled her over. He immediately called 911 and then cradled her head in his arms.
Deputy Nourse also said that although it was an overcast night and very dark, and although only half an hour had transpired between petitioner’s reported arrival at his property and Nourse’s arrival there, petitioner was able to take the deputy on a detailed tour of the crime scene. Petitioner knew that Pamela’s pants were lying next to the generator, and that they had not come off easily, telling the deputy, “trust me on this.” He knew that her underwear was inside the camper. He knew that her blood was inside the camper on the pillow. He knew that there was “blood on rocks up against the hill” (referring to the rough, upward-sloping terrain to the southwest of the crime scene). He knew that a bloodstained paving stone had been thrown “over the side of the hill” (referring to the rough, downward-sloping terrain to the north of the crime scene). He theorized about what Pamela was doing when her murderer arrived, where the murderer confronted Pamela, and what she did in her defense. And he surmised that the murderer had used a cinder block to kill Pamela. He also told the deputy: “[A]ll the evidence that relates to this case I already touched and moved trying to figure out how this whole thing happened.”
Deputy Nourse described petitioner’s demeanor as “very calm, cool, [and] collected,” but occasionally petitioner would fall to his knees crying, after which he would get back up and continue talking. To the deputy, it seemed as if petitioner was speaking “like he had rehearsed or was reading from a script.” Petitioner’s dogs barked, growled, and snarled at Deputy Nourse. Petitioner remarked that the dogs had failed to protect his wife from her murderer. He did not report anything missing from the premises.
Deputy Nourse checked Pamela’s body. It was neither warm nor cold. Her arm was pliable. Her blood was still wet, bright red, and in a puddle; it had not coagulated or soaked into the sandy soil. She appeared to have just died.
*954Sheriff’s investigators secured the crime scene. In the morning, they conducted a thorough investigation. Pamela was lying on her back on the ground outside the camper, covered with a sleeping bag. She was naked from the waist down, except for her socks. Her head was crushed, an eye hanging out. A bloody cinder block was lying near her head, and blood spatters were on the ground. Petitioner had scattered bloodstains on his pants and shoes. From a study of all the footprints and tire tracks that were discernible in the soft sand and gravel, it did not appear that anyone had been present except petitioner, Pamela, and the investigators.
Investigators found a note in Pamela’s purse in which petitioner proposed a division of their assets and personal property. Petitioner and his wife had been having financial and marital difficulties, and both had sexual relationships outside the marriage.
DNA testing established that the bloodstains on petitioner’s pants and shoes belonged to Pamela. A criminalist determined that the stains were from blood spatter, not from drips or contact, indicating that the blood hit petitioner’s pants and shoes when Pamela’s skull was smashed.
An autopsy determined that Pamela had been strangled. The strangling was sufficient by itself to cause her death. In addition, her skull was smashed. That injury was also sufficient to cause her death. There was no evidence of sexual assault.
The pathologist who performed the autopsy severed some of Pamela’s fingertips from her body for testing. Daniel Gregonis, a criminalist, later examined the severed fingertips under a stereomicroscope and noticed blue cotton fibers wedged deep in a crack of a broken fingernail. A broken fragment apparently tom from the same fingernail was found on the ground at the crime scene, suggesting that the fingernail broke during Pamela’s straggle with her assailant. The criminalist examined the-blue cotton shirt petitioner wore on the night of the murder, and he concluded that the blue shirt fibers were indistinguishable from the fibers caught in the crack in Pamela’s broken fingernail. The same fibers were not, however, mentioned in the report of Craig Ogino, another criminalist who testified for the prosecution at trial and who may have examined the same fingertips.
The timeclock at petitioner’s work indicated that he had left there at 11:03 p.m. on the night of the murder. A few weeks after Pamela’s murder, a sheriff’s homicide investigator went to petitioner’s place of employment. The investigator left petitioner’s workplace at 11:03 p.m., walked to his car, and then drove at the speed of traffic (60-70 miles per hour). He arrived at *955petitioner’s residence at 11:47 p.m. If on the night of the murder petitioner also arrived home at 11:47 p.m., then he had been at home for 11 minutes when he called 911.
A homicide detective interviewed petitioner on several occasions after Pamela’s murder. Petitioner’s statements were generally consistent with what he had earlier told Deputy Nourse, who responded to petitioner’s 911 call.
B. Petitioner’s Trial and Conviction
Petitioner was charged with murder. (Pen. Code, § 187.) His first trial ended in a mistrial after the jury was unable to reach a verdict. His second trial was aborted before a jury was selected, when the trial court recused itself. His third trial, like his first, ended in a mistrial after the jury could not reach a verdict.
At petitioner’s fourth trial, the prosecution presented the evidence described in part I.A., ante. In addition, the prosecution for the first time presented expert testimony by a forensic dentist, Dr. Norman D. Sperber. Dr. Sperber described a unique feature of petitioner’s lower teeth: The lower right canine tooth was out of alignment with the other teeth and had not emerged fully from the gum. Dr. Sperber testified, based solely on his experience as a practicing dentist, and expressly without the benefit of any scientific studies, that “it might be one or two or less” out of a hundred people who would have petitioner’s dental irregularity. After visually comparing a photograph of an indistinct crescent-shaped lesion on murder victim Pamela’s hand to a model of petitioner’s lower teeth, Dr. Sperber stated his opinion that the lesion was a human bite mark, and that petitioner’s unusual dentition was “consistent with” the bite mark. In the photograph, the lesion appears only as a reddening and bruising of the surface tissues; Pamela’s skin is not broken, and it is difficult to identify individual marks that might be tooth marks.
The defense at trial presented testimony from several witnesses who said that both petitioner and Pamela seemed “fine” and “normal” on the day of the murder, although Pamela’s brother testified that Pamela had told him she and petitioner had been arguing.
In addition, the defense presented evidence that Pamela had already been dead for some time when petitioner arrived home on the night of the murder. Dr. Griffith Thomas, a forensic pathologist, testified about the factors from which the time of death can be determined. In his view, the time of Pamela’s death was uncertain, but many of her contusions occurred several hours before she died.
*956In regard to the supposed bite mark on Pamela’s hand, Dr. Gregory S. Golden, an expert in forensic dentistry, testified for the defense that in a brief review of 15 “study models” of teeth in his office, he found five models that were “consistent with” the mark. In his view, the bite-mark evidence was inconclusive and should be disregarded, in part because of the angular distortion in the photograph of the mark. On cross-examination, Dr. Golden agreed with Dr. Sperber’s estimate that petitioner’s displaced canine tooth occurred in only about 2 percent of the general population.
Finally, defense witness Dean M. Gialamas, a senior criminalist with the Los Angeles County Sheriff’s Department, testified that the scattered bloodstains on the clothes petitioner was wearing on the night of Pamela’s murder were contact stains, not spatter stains, and they were more consistent with petitioner’s story that he cradled Pamela’s dead body than with the prosecution’s theory that petitioner was the killer.
The jury found petitioner guilty of first degree murder of his wife, Pamela. (Pen. Code, § 187, subd. (a).) The trial court sentenced petitioner to 25 years to life in state prison. The judgment was affirmed by the Court of Appeal.
C. Habeas Corpus Proceeding
In 2007, petitioner sought habeas corpus relief in the San Bernardino County Superior Court, asserting that his 1997 murder conviction was based on false evidence, and that new evidence unerringly established his innocence. Among the voluminous exhibits supporting the petition was a declaration from Forensic Dentist Sperber, discussing his statement at trial that only 1 or 2 percent of the population had petitioner’s dental irregularity. Dr. Sperber’s declaration stated: “These percentages were based on my own experience and were not scientifically accurate.” Concerning the lesion on murder victim Pamela’s hand, the declaration added: “With the benefit of all of the photographs [of the crime scene and Pamela’s injuries], and with my added experience, I would not now testify as I did in 1997,” and “I cannot now say with certainty that the injury on the victim’s hand is a human bite mark injury.”
Also supporting the petition were declarations and reports by other experts in forensic dentistry. The declaration of Dr. Golden, who had testified for the defense at trial, said that he had “enlarged the image [of murder victim Pamela’s hand lesion] to life-size,” and that after comparing the enlarged image to petitioner’s teeth, he “would tend to exclude [petitioner] as the suspected biter.” The declaration of Dr. Charles M. Bowers described the process by which he and Dr. Raymond J. Johansen had removed angular distortion from the photograph of Pamela’s hand lesion, thereby permitting a *957more accurate comparison between the hand lesion and petitioner’s lower teeth. According to Dr. Bowers, “[t]he new scientific methods demonstrably contradict the conclusion at trial that [petitioner] could not be ruled out as a suspected biter.” Dr. Bowers also criticized the methodology used by Dr. Sperber at petitioner’s trial. In an earlier written report, Dr. Bowers said that there is “significant doubt that the hand injury is even a bitemark.”
The superior court issued an order to show cause and held an evidentiary hearing on the habeas corpus petition. The court heard testimony from Dr. Sperber (the prosecution’s forensic dentistry witness at trial) and from the other dental experts who had submitted declarations and reports in support of habeas corpus relief. Dr. Sperber discussed the points he had made in his posttrial declaration. He also described the angular distortion in the photograph of the lesion on Pamela’s hand, and he examined photographs depicting other lesions on Pamela’s body. Referring to the photograph of the lesion on Pamela’s hand, he said: “I don’t know for sure that . . . that photograph depicts a bite mark.” Dr. Sperber added: “My opinion today is that [petitioner’s] teeth ... are not consistent with the lesion on the hand.” Dr. Golden likewise confirmed the points he had made in his declaration. He also described the availability of new computer technology allowing him to remove angular distortion from photographs. He concluded that the lesion on Pamela’s hand might have been from a dogbite or some other source; in any case, he “would tend to rule out Mr. Richards ... as the suspected biter.”
Drs. Bowers and Johansen testified in detail about the process by which they digitally altered the photograph of Pamela’s hand to remove angular distortion. Drs. Bowers and Johansen, experts in this process, have published articles and a book describing the use of this process in forensic dentistry. The technique was first used in 1996 or 1997 by a single dentist in Canada, after which Drs. Bowers and Johansen further developed the technique, which has since become accepted in the field of forensic dentistry.
Dr. Johansen said that he removed the angular distortion from the photograph of Pamela’s hand and then compared the corrected photograph to “overlays” depicting petitioner’s lower and upper teeth. He also examined photographs of wire-mesh fencing material found on the ground near Pamela’s body. He concluded that petitioner’s lower teeth did not match the lesion on Pamela’s hand. The upper teeth matched the lesion in some places, but not others. Dr. Johansen could not exclude petitioner’s teeth as a possible source of the lesion, but in his opinion it was just as likely that the indistinct lesion was caused by the fencing material as it was by petitioner’s teeth. But Dr. Johansen conceded on cross-examination that he had removed angular distortion from the same photograph in 2000, concluding at that time that the lesion was a human bite mark that was consistent with petitioner’s teeth.
*958Dr. Bowers testified that he likewise removed the angular distortion from the photograph of murder victim Pamela’s hand and compared the corrected photograph to an “overlay” of petitioner’s lower teeth. He found no match. After examining photographs of other lesions on Pamela’s body, Dr. Bowers doubted whether the hand lesion was a human bite mark.
Other evidence presented in support of the habeas corpus petition indicated that the blue fibers found embedded in a crack of Pamela’s broken fingernail might not have been present at the time of the autopsy. No blue fibers were visible in a still photograph that was taken during the autopsy, nor were they visible when the autopsy photograph was digitally altered to increase color saturation. By contrast, the blue fibers were visible in a still photograph taken from a video recording of the postautopsy testing of Pamela’s fingertips.
In addition, a two-centimeter-long hair was found under one of Pamela’s long artificial fingernails. A new type of DNA testing revealed that this hair did not come from Pamela or from petitioner, and the district attorney did not dispute those DNA test results. Patricia Zajac, a professor of criminal justice at California State University, East Bay and a criminalist, reviewed the laboratory notes and other records in the case. She testified that in her opinion the hair was not “historical,” by which she meant that in her view its placement was somehow related to the violence that resulted in Pamela’s death. Professor Zajac also testified that although the hair had a mature root and had fallen out naturally (or was ready to do so), that fact did not undermine her conclusion that the hair under Pamela’s fingernail was related to the crime.
Finally, a new type of DNA testing indicated that the bloodstained paving stone that—along with a cinder block—may have been used to smash Pamela’s skull had trace DNA from one or more male persons, none of whom was petitioner. Pamela’s DNA, however, was the primary DNA on the paving stone. The ratio of the male DNA to Pamela’s DNA was 1 to 10 (in one location) and one to six (in another location). Again, the district attorney did not dispute those DNA test results. In addition, the male DNA was located on the paving stone in a place near where Daniel Gregonis, a criminalist in the sheriff’s department, had earlier expected he might find the DNA of the murderer.
The district attorney called Gregonis as a witness. Gregonis and another criminalist, Craig Ogino, had examined “scrapings” from Pamela’s fingernails. These scrapings included a dark animal hair and also the two-centimeter-long human hair that, according to petitioner’s DNA testing, did not come from either Pamela or from petitioner. Because of the length of Pamela’s artificial fingernails, Gregonis was of the view that the human hair *959might have been lodged under Pamela’s fingernail for a long time without being noticed, and he therefore concluded that the hair might not have been related to the crime. The parties stipulated that criminalist Ogino would testify along similar lines.
Gregonis also discussed the male DNA found on the bloodstained paving stone that might have been used to kill Pamela. He said that the male DNA was present in an amount that was consistent with contamination that could have occurred in the courtroom, although it also could have been already present on the stone when the stone was stained with Pamela’s blood.
The superior court granted habeas corpus relief, concluding that the new evidence pointed unerringly to petitioner’s innocence. In reaching this conclusion, however, the superior court made very few specific findings of fact. The court ordered that petitioner be remanded for a new trial, but it stayed its decision for 15 court days to allow the district attorney to appeal.
To preserve the status quo, the Court of Appeal stayed proceedings to retry petitioner. After briefing and oral argument, the Court of Appeal vacated the superior court’s order granting the petition for a writ of habeas corpus. With regard to petitioner’s false evidence claim (challenging the trial testimony of Dr. Sperber regarding the alleged bite mark), the Court of Appeal held that a habeas corpus petitioner does not establish false evidence under Penal Code section 1473’s subdivision (b) merely by presenting new expert testimony on how to interpret the evidence offered at the trial. Accordingly, the Court of Appeal concluded that all petitioner’s claims were, in effect, new evidence claims, and that the unerring innocence standard applicable to such claims therefore applied to all his claims. The court reviewed the evidence that petitioner had offered in support of habeas corpus relief and concluded that petitioner’s evidence failed to establish unerring innocence.
We granted petitioner’s petition for review. His claims are addressed below.
II
Penal Code section 1473, subdivision (b) provides in relevant part; “A writ of habeas corpus may be prosecuted for, but not limited to, the following reasons: [f] (1) False evidence that is substantially material or probative on the issue of guilt or punishment was introduced against a person at any hearing or trial relating to his incarceration . . . .” (Italics added.) In addition, this court’s decisions hold that habeas corpus relief is appropriate if the petitioner presents new evidence that unerringly establishes innocence. With regard to new evidence, we have said: “It is not sufficient that the *960evidence might have weakened the prosecution case or presented a more difficult question for the judge or jury. [Citations.] ‘[A] criminal judgment may be collaterally attacked on the basis of “newly discovered” evidence only if the “new” evidence casts fundamental doubt on the accuracy and reliability of the proceedings. At the guilt phase, such evidence, if credited, must undermine the entire prosecution case and point unerringly to innocence or reduced culpability I [Citation.]” (In re Clark, supra, 5 Cal.4th at p. 766, italics added; see In re Lawley (2008) 42 Cal.4th 1231, 1238-1241 [74 Cal.Rptr.3d 92, 179 P.3d 891].) Here, petitioner asserts both theories of relief; that is, he asserts that his conviction was based on false evidence, and that new evidence unerringly establishes his innocence.
Our standard of review is de novo with respect to questions of law and the application of the law to the facts. We accept as final the superior court’s resolution of pure questions of fact if they are supported by substantial evidence. (In re Collins (2001) 86 Cal.App.4th 1176, 1181 [104 Cal.Rptr.2d 108].) Here, the superior court made very few specific findings of fact regarding the evidence, and none that is significant to the main part of our analysis. Accordingly, our review here is de novo.
A. False Evidence
1. Legal principles
As noted, Penal Code section 1473’s subdivision (b) provides for habeas corpus relief if “[f]alse evidence that is substantially material or probative on the issue of guilt or punishment was introduced against a person at any hearing or trial relating to his incarceration . . . .” Before 1975, it was much more difficult for a convict to obtain habeas corpus relief on false evidence grounds, requiring proof that the prosecution knowingly relied on perjured testimony. As explained by the Court of Appeal in In re Wright (1978) 78 Cal.App.3d 788 [144 Cal.Rptr. 535] (Wright): “Prior to the 1975 amendment to Penal Code section 1473, the rule was clear that to obtain habeas corpus relief on the ground of perjured testimony, the petitioner was required to establish by a preponderance of the evidence: (1) that perjured testimony was adduced at his trial, (2) that this was known to a representative of the state, and (3) that the perjured testimony may have affected the outcome of the trial. (In re Imbler [(1963)] 60 Cal.2d [554,] 560 [35 Cal.Rptr. 293, 387 P.2d 6]; Napue v. Illinois [(1959)] 360 U.S. 264, 269, 272 [3 L.Ed.2d 1217, 79 S.Ct. 1173]; see Witkin, Cal. Criminal Procedure (1975 supp.) § 804, p. 866.)” (Wright, at p. 807.)
The Court of Appeal in Wright, supra, 78 Cal.App.3d 788, related the key changes that the 1975 amendment had effected: “Under Penal Code *961section 1473, subdivision (b)(1) ... it is not required that perjury be proved. A showing of ‘false evidence’ is sufficient. Additionally, it is no longer necessary to show a representative of the state knew the testimony was false. (Pen. Code, § 1473, subd. (c).)” (Wright, at p. 809, fn. 5.) Wright relied on and quoted a committee analysis of the Assembly bill that led to the 1975 amendment to Penal Code section 1473: “ ‘The key distinction between [prior case law] . . . , on the one hand, and this bill, on the other, is that the former all involve prosecutorial misconduct, while, under the bill, the [proposed prejudice] standard would apply whether or not use of false evidence by the prosecution had been knowing, negligent, inadvertent, or totally without fault.' " (Wright, at pp. 810-811, fn. 6, original italics.)
The Court of Appeal in Wright, supra, 78 Cal.App.3d 788, also noted an important way in which the law remained unchanged after the 1975 amendment: “[T]he requirement under the preexisting law that the petitioner show the false evidence may have affected the outcome of his trial was not eliminated or changed by the 1975 amendment to Penal Code section 1473 and is still required for relief under the amended statute: ‘[f]alse evidence that is substantially material or probative on the issue of guilt or punishment’ means false evidence of such significance that it may have affected the outcome of the trial . . . .” (Wright, at pp. 808-809, italics added.) Wright further explained that this required showing of prejudice was the same as the “reasonably probable” test (see People v. Watson (1956) 46 Cal.2d 818, 836 [299 P.2d 243]) that applies in other instances of state law error. (Wright, at p. 812.) As we said in In re Roberts (2003) 29 Cal.4th 726, 742 [128 Cal.Rptr.2d 762, 60 P.3d 165): “False evidence is ‘substantially material or probative’ [citation] ‘if there is a “reasonable probability” that, had it not been introduced, the result would have been different. [Citation.]’ [Citation.]”
Thus, under the present version of section 1473 it does not matter why evidence is false or whether any party to the proceeding knew it was false. So long as some piece of evidence at trial was actually false, and so long as it is reasonably probable that without that evidence the verdict would have been different, habeas corpus relief is appropriate.
We summarized those principles in In re Hall (1981) 30 Cal.3d 408, 424 [179 Cal.Rptr. 223, 637 P.2d 690]: “The new law requires only that the evidence be ‘false’ and ‘substantially material or probative on the issue of guilt or punishment’ ([Pen. Code, § 1473], subd. (b)(1)); there is no longer any obligation to show that the testimony was perjured or that the prosecutor or his agents were aware of the impropriety. (In re Wright, supra, 78 Cal.App.3d 788, 807-808.) [f] In this case, the trial testimony of [two key witnesses] identifying petitioner as the killer apparently was false, albeit unintentionally so. As it was virtually the only damning evidence against *962petitioner, that testimony clearly satisfies the statute’s test of materiality. Accordingly, ... we hold that false evidence was introduced against petitioner, and that issuance of the writ is justified on that ground . . . .” (Italics added.)
2. Expert opinion testimony
Here, we consider how to apply the above discussed principles in the context of an expert witness’s opinion testimony. Expert opinion is qualitatively different from eyewitness testimony and from physical evidence. Expert witnesses are by definition witnesses with “special knowledge, skill, experience, training, or education” in a particular field (Evid. Code, § 720), and they testify “in the form of an opinion” (id., § 801, italics added). The word “opinion” implies a subjective component to expert testimony. Moreover, expert witnesses may only testify about matters that are “beyond common experience” (id., § 801, subd. (a)), and such witnesses very often testify regarding matters that lie at the frontier of human knowledge about a given subject. It is not surprising that in the latter circumstance, an expert’s opinion may rely to some extent on evolving theories, assumptions, or methods. Moreover, technical limitations may inhibit an expert witness’s analysis of the available data. Thus, it is conceivable—even reasonable—that an expert witness’s opinion may change over time without that change implying any lack of integrity on the expert’s part.1
For example, research and scientific breakthroughs may have altered the generally accepted hypotheses of the expert’s field, or new technologies may have opened up more accurate methods for resolving disputed questions. In either situation, it may be that expert witness opinion testimony in a criminal trial is later proved to be objectively untrue. If, with hindsight, a critical component of the prosecution’s case is objectively untrue, then the validity of any resulting guilt finding is called into question. It does not matter why the critical evidence was untrue; regardless of why it was untrue, the fact that it was untrue, coupled with the fact that it affected the outcome of the trial, casts a doubt over the verdict of guilt. In such circumstances, the law places the importance of integrity in criminal trials above the public’s interest in the finality of the judgment. (In re Hall, supra, 30 Cal.3d at p. 424.)
Given, on the one hand, the subjective component of expert opinion testimony, and, on the other hand, the possibility that advances in science and technology might prove an earlier-held opinion to be objectively untrue, it is critical to define what precisely is meant by “false” when the false evidence standard of Penal Code section 1473 is applied to expert opinion testimony.
*963When an expert witness gives an opinion at trial and later simply has second thoughts about the matter, without any significant advance having occurred in the witness’s field of expertise or in the available technology, it. would not be accurate to say that the witness’s opinion at trial was false. Rather, in that situation there would be no reason to value the later opinion over the earlier. Therefore, one does not establish false evidence merely by presenting evidence that an expert witness has recanted the opinion testimony given at trial. Likewise, when new expert opinion testimony is offered that criticizes or casts doubt on opinion testimony given at trial, one has not necessarily established that the opinion at trial was false. Rather, in that situation one has merely demonstrated the subjective component of expert opinion testimony.
When, however, there has been a generally accepted and relevant advance in the witness’s field of expertise, or when a widely accepted new technology has allowed experts to reach an objectively more accurate conclusion, a strong reason may exist for valuing a later opinion over an earlier opinion. If, and only if, a preponderance of the evidence shows that an expert opinion stated at trial was objectively untrue, the false evidence standard applies.2 In that narrow circumstance, if it is reasonably probable that the invalid opinion given at trial affected the verdict, then habeas corpus relief is appropriate.
With these principles in mind, we turn to the facts surrounding the issue here.
3. Dr. Sperber’s opinion at trial
In regard to his false evidence claim, petitioner focuses on Dr. Sperber’s expert opinion given at trial that the arrangement of petitioner’s lower teeth *964occurs in only 1 or 2 percent of the general population, and that his lower teeth may have been the source of the lesion on Pamela’s hand.3 At the evidentiary hearing in the superior court, petitioner presented the testimony of Dr. Sperber himself and three other experts in forensic dentistry. Discussing his trial testimony that “it might be one or two or less” out of a hundred people who would have petitioner’s specific dental irregularity, Dr. Sperber said that he had not based his testimony on any scientific studies, and that without such studies, he should not have testified to statistical percentages. With regard to the lesion on Pamela’s hand, Dr. Sperber and three other experts cast doubt on Dr. Sperber’s trial testimony. These experts were uncertain that the lesion was a human bite mark; if it was, they all—including Dr. Sperber—agreed that petitioner’s teeth did not definitively match the lesion.
Three of these experts—Drs. Golden, Bowers, and Johansen—relied on the availability of new technology in reaching their opinions. Using a computer program, these experts were able to remove angular distortion from the photograph of the lesion on murder victim Pamela’s hand, thus artificially generating an accurately sized front-facing image of the lesion. The experts then laid an image of petitioner’s lower teeth over the corrected image of the lesion, and they found no match, although they could not definitively rule out petitioner’s teeth as a possible source of the lesion.
As explained earlier (pp. 962-963, ante), Dr. Sperber’s posttrial change of view regarding the statistical prevalence of petitioner’s abnormal dentition and regarding the cause of the lesion on murder victim Pamela’s hand does not by itself establish that his opinion offered at trial was “false evidence” under Penal Code section 1473, subdivision (b). In the habeas corpus proceedings, Dr. Sperber did not testify that he relied on new technology in revising his views about the lesion. When an expert witness merely changes an earlier-given opinion, without relying on any significant advance in the witness’s field of expertise or in the available technology, no reason exists to value the later opinion over the earlier. Likewise, the fact here that other dental experts disagree with or are critical of the opinion Dr. Sperber gave at trial does not by itself establish that his trial opinion was false. As noted (pp. 962-963, ante), opinion testimony often includes a subjective component, and good faith disagreements among credible experts are commonplace. That point is particularly relevant here, where the question at issue was an attempt to surmise the cause of an indistinct reddening and bruising of the murder victim’s skin.
*965But the opinion Dr. Sperber offered at trial could qualify as “false evidence” for purposes of Penal Code section 1473’s subdivision (b) if, for example, a generally recognized and relevant advance in science or technology proved under the preponderance of the evidence standard that the trial opinion was objectively untrue. (See p. 963, ante.)
The district attorney does not contend that the technology to remove angular distortion from photographs existed at petitioner’s 1997 murder trial, and that the defense therefore should have used such technology at trial to impeach Dr. Sperber. (See People v. Marshall (1996) 13 Cal.4th 799, 830-831 [55 Cal.Rptr.2d 347, 919 P.2d 1280] [“We conclude defendant waived his claim that his conviction was based on false testimony by failing to raise it at trial when the falseness of [the] testimony was well known to him . . . .”].) In addition, testimony at the 2007 habeas corpus evidentiary hearing established that the technology to remove such distortion did not exist.4 In light of that testimony, we conclude that this case involves a generally recognized and relevant advance in technology. Here, however, the new technology has not proved that any portion of Dr. Sperber’s trial testimony was objectively untrue.
Dr. Sperber testified at the 1997 trial that petitioner’s dental irregularity “might” occur in 2 percent or less of the general population, but he added that this estimate was not based on any scientific studies. The availability of new technology to remove angular distortion from photographs has not proved that Dr. Sperber’s estimate regarding the prevalence of petitioner’s abnormal dentition was objectively untrue. Moreover, Dr. Sperber’s testimony at the habeas corpus proceeding merely reiterated what he had said at trial, that the percentage was an estimate that was not based on any scientific studies.
Dr. Sperber further testified at trial that the lesion on murder victim Pamela’s hand was a human bite mark, and that he could not rule out petitioner’s teeth as a possible source of the mark. With the benefit of new technology, petitioner’s experts at the habeas corpus evidentiary hearing shed doubt on those conclusions, but even with the new technology, these experts still could not definitively rule out petitioner’s teeth as a possible *966source of the mark.5 Dr. Johansen, for example, could not exclude petitioner’s teeth as a possible source of the mark, but in his opinion it was just as likely that the indistinct lesion was caused by the fencing material as it was by petitioner’s teeth. The other doctors found no match to petitioner’s teeth, but they also did not absolutely rule out petitioner’s teeth as a possible source of the mark. Petitioner’s habeas corpus evidence at most calls into question Dr. Sperber’s opinion at trial that petitioner’s teeth could have been the source of the mark, but it has not proved that opinion to be objectively untrue. Hence, Dr. Sperber’s trial opinion is not “false evidence” for purposes of section 1473, subdivision (b).
Because petitioner failed to establish that any of the evidence offered at his 1997 trial was objectively false, the false evidence provision of Penal Code section 1473’s subdivision (b) does not apply here, and petitioner is not entitled to habeas corpus relief on that ground. Nevertheless, the question remains whether he is entitled to habeas corpus relief on the basis of newly discovered evidence.
*967B. New Evidence
As noted, habeas corpus relief is appropriate if the petitioner presents new evidence that “ ‘point[s] unerringly to innocence or reduced culpability.’ ” (In re Clark, supra, 5 Cal.4th at p. 766.) Petitioner argues that the evidence he presented at the evidentiary hearing meets this standard. We disagree, as discussed below.
1. Lesion on Pamela’s hand
As discussed in regard to petitioner’s false evidence claim (pp. 963-964, ante), Drs. Golden, Bowers, and Johansen testified at the evidentiary hearing that by removing the angular distortion from the photograph of the lesion on murder victim Pamela’s hand, they concluded that petitioner’s teeth did not match the lesion, and that the lesion might not even be a bite mark. This evidence tended to undermine Dr. Sperber’s opinion at the 1997 trial that the lesion was a bite mark, and that petitioner’s teeth could not be ruled out as a possible source of the mark. This new evidence, however, does not “ ‘point unerringly to innocence or reduced culpability’ ” (In re Clark, supra, 5 Cal.4th at p. 766), as we explain below.
The case against petitioner was strong. When Pamela’s murder took place, petitioner and Pamela were in the process of ending their marriage, and Pamela planned on leaving petitioner. No other motive (such as robbery or rape) appears for Pamela’s murder. The remote property where the couple lived was guarded by several dogs that were hostile to strangers, so it is unlikely that a person other than petitioner (who was familiar to the dogs) would have had access to the property. Footprints and tire tracks in the soft ground at the couple’s property indicated that no one other than petitioner, Pamela, and the sheriff’s investigators had been present on the night of the murder. When the deputy sheriff responding to petitioner’s 911 telephone call arrived at the murder scene at 12:30 a.m., petitioner’s demeanor seemed “rehearsed,” and petitioner knew an unusual amount of detail about the crime scene, despite the darkness. (See p. 953, ante.)
At that time, petitioner told the deputy that the battery in the camper had lost its charge, that he did not turn on the generator, and that he had no light. Yet petitioner was able to take the deputy on a detailed tour of the crime scene. He knew that Pamela’s pants were lying next to the generator. He knew that her underwear was inside the camper. He knew that her blood was inside the camper on the pillow. He knew that there was “blood on rocks up against the hill.” He knew that there was a bloodstained paving stone that had been thrown “over the side of the hill.” He also theorized about what Pamela was doing when her murderer arrived, where the murderer confronted *968Pamela, and what she did in her defense. And he surmised that the murderer had used a cinder block to kill Pamela. (See p. 953, ante.)
Furthermore, Pamela’s artificial fingernail was broken (apparently in her struggle with her assailant), and fibers matching petitioner’s blue cotton shirt were found wedged in the crack of the broken fingernail. Also, petitioner had Pamela’s bloodstains on his pants and shoes—stains that in the opinion of the prosecution’s expert witness were from blood spatter, not from drips or contact (indicating that petitioner was present when Pamela’s skull was smashed). (See p. 954, ante)
All of that evidence pointed persuasively to petitioner’s guilt. Dr. Sperber’s trial testimony was also evidence pointing to petitioner’s guilt, although it was hardly conclusive. We cannot say that petitioner has “unerringly” established his innocence (In re Clark, supra, 5 Cal.4th at p. 766) merely by casting doubt on Dr. Sperber’s conclusions, when he has not undermined the other evidence establishing his guilt.
2. Blue fibers
At the 1997 trial, the prosecution’s evidence showed that murder victim Pamela’s artificial fingernail was broken (apparently in her struggle with her assailant), and that wedged in the crack of the broken fingernail were blue fibers indistinguishable from those in the blue cotton shirt that petitioner was wearing on the night of the murder. Petitioner’s evidence presented at the 2007 habeas corpus evidentiary hearing indicates that the blue fibers embedded in the crack in Pamela’s fingernail may have found their way into that crack after the crime, not before. This is not new evidence, however. (See In re Hall, supra, 30 Cal.3d at p. 420 [a habeas corpus petitioner must present evidence that was unavailable at trial].) It is based on an autopsy photograph that existed at the time of petitioner’s trial. In seeking habeas corpus relief, petitioner has merely made a high resolution scan of that photograph and increased the color saturation. Moreover, the evidence is far from conclusive. Petitioner’s evidence shows that the fibers, which were visible in a video made after the autopsy, were not visible in the still photograph taken during the autopsy. That fact, however, does not mean that the fibers were not present. Petitioner offered no evidence regarding the technology used to make the original autopsy photograph (in which the fibers were not visible). By contrast, the video (in which the fibers were visible) was made by attaching a video recorder to a microscope that magnified Pamela’s fingers between 10 and 100 times. The technology by which the earlier photograph was made may not have resulted in a photograph with sufficient detail to record the presence of the microscopic blue fibers, and a high resolution scan of a *969photograph cannot improve the limitations of the original. Thus, the nonappearance of the blue fibers in the earlier photograph does not conclusively establish their absence from Pamela’s fingernail.
Moreover, prosecution criminalist Daniel Gregonis, who detected the blue fibers, testified at trial that they were embedded deep in the crack of Pamela’s broken fingernail. Therefore, it may be that the blue fibers became visible only after Gregonis opened the crack in the fingernail and removed some of the fibers for closer inspection.
Petitioner also presented evidence that prosecution trial witness Craig Ogino, who examined scrapings from Pamela’s fingertips, and who may also have examined Pamela’s fingers, did not mention the blue fibers in his report. But this evidence is inconclusive. The blue fibers were first discovered by criminalist Gregonis when he used a microscope to examine Pamela’s fingertips. Ogino may not have examined the fingertips, and if he did, he may have done so without the aid of a microscope, and therefore he may have overlooked the fibers.
As noted (pp. 967-968, ante), the trial record includes significant evidence pointing persuasively to petitioner’s guilt, including evidence of motive and opportunity, evidence that petitioner knew details of the crime scene that only the murderer could know, and evidence of blood spatter on petitioner’s clothes and shoes, indicating that he was present when Pamela’s head was smashed. Given all that evidence, we cannot say that petitioner has “unerringly” established his innocence (In re Clark, supra, 5 Cal.4th at p. 766) simply by presenting evidence that the blue fibers caught in a crack in Pamela’s broken fingernail may have come from a source other than petitioner.
3. Two-centimeter-long hair
A similar analysis applies to the two-centimeter-long human hair that investigators found under Pamela’s long artificial fingernail. Petitioner’s evidence at the habeas corpus evidentiary hearing indicates that this hair belonged to some third person, not to Pamela or to petitioner. That fact, however, does not point unerringly to petitioner’s innocence. The hair might have been from someone Pamela encountered at the restaurant where she worked or elsewhere. Because Pamela had “extended” artificial fingernails, the hair might have been lodged under the fingernail for a day or more without being noticed. In light of the significant evidence pointing to petitioner’s guilt (pp. 967-968, ante), petitioner has not established “ ‘unerring[] . . . innocence’ ” (In re Clark, supra, 5 Cal.4th at p. 766) simply by showing that the piece of hair under Pamela’s fingernail was neither hers nor petitioner’s.
*9704. Male DNA on paving stone
Petitioner has not proved unerring innocence by presenting evidence at the 2007 habeas corpus proceeding that some of the DNA on the paving stone that might have been used to kill Pamela came from one or more men other than petitioner.
At the 1997 trial, prosecution criminalist Gregonis testified that the cinder block was one of the murder weapons, but that the paving stone might also have been used as a weapon. His laboratory notes indicated a specific location on the paving stone where he expected to find the murderer’s DNA. After the trial, petitioner’s expert found traces of male DNA not belonging to petitioner in the approximate location on the paving stone that Gregonis had indicated in his notes. But Gregonis testified at the habeas corpus evidentiary hearing that the DNA in question was a small amount that could have come from contamination of the evidence during courtroom handling. We conclude the presence on the bloodstained paving stone of trace DNA that was not from murder victim Pamela or from petitioner does not “ ‘point unerringly to [petitioner’s] innocence’ ” (In re Clark, supra, 5 Cal.4th at p. 766), in light of all the evidence supporting petitioner’s conviction of his wife’s murder. (Pp. 967-968, ante.)
5. Cumulative effect of new evidence
Finally, even when all the new evidence is considered together, we conclude that it does not “ ‘point unerringly to [petitioner’s] innocence.’ ” (In re Clark, supra, 5 Cal.4th at p. 766.) The evidence that petitioner was the actual murderer remains strong.
Because petitioner has failed to establish that any of the evidence offered at his 1997 trial was false (Pen. Code, § 1473, subd. (b)), and because his newly discovered evidence does not “ ‘point unerringly to innocence or reduced culpability’ ” (In re Clark, supra, 5 Cal.4th at p. 766), the superior court erred in granting habeas corpus relief, and the Court of Appeal was correct to vacate the superior court’s order, and to direct that court to deny the petition for a writ of habeas corpus.
Disposition
The judgment of the Court of Appeal is affirmed
Cantil-Sakauye, C. J., Baxter, J., and Corrigan, J., concurred.
Our discussion here presumes good faith expert opinion testimony. Perjury is always “[f]alse evidence” for purposes of Penal Code section 1473, subdivision (b).
The dissent asserts that a habeas corpus petitioner need not show that the expert’s ultimate conclusion was untrue; rather, the petitioner need only undermine the analytical basis of that conclusion. (Dis. opn., post, at pp. 972-976.) Thus, according to the dissent, petitioner here need not show that Dr. Sperber’s conclusion was wrong regarding the source of the lesion on murder victim Pamela’s hand; rather, petitioner need only show that Dr. Sperber relied on a false assumption to reach that conclusion. (Ibid.)
The dissent’s interpretation of Penal Code section 1473’s subdivision (b) would permit habeas corpus relief even when the petitioner’s own experts at the habeas corpus proceeding fully agreed with the conclusion of the prosecution’s trial expert. If, for example, petitioner’s experts agreed with Dr. Sperber that the lesion at issue here was a bite mark, and also that it was consistent with petitioner’s unusual dentition, but those experts had a different scientific basis for reaching their conclusion, then the dissent would apparently permit habeas corpus relief on the ground that Dr. Sperber’s scientific basis was subject to criticism, notwithstanding petitioner’s experts’ confirmation of Dr. Sperber’s conclusion. We doubt that the Legislature had that in mind when it used the term “false evidence” in Penal Code section 1473’s subdivision (b).
Petitioner also asserts that the blue fibers that prosecution criminalist Gregonis found jammed into a crack in murder victim Pamela’s fingernail constituted false evidence, insinuating that they were planted there by Gregonis. The superior court, however, made a specific factual finding—at the close of the evidentiary hearing on petitioner’s habeas corpus claims— rejecting this false evidence claim, and that finding has ample support in the record before us.
Dr. Sperber testified at the hearing that the technology in question was not, to his knowledge, available in 1997. He acknowledged that there may have been some initial studies, but he said that forensic dentists were not using the technology. Dr. Golden agreed with Dr. Sperber on this point. Drs. Bowers and Johansen testified that there were a few articles in the late 1990’s that led to their use of the technique starting in 2000. In addition, Dr. Bowers’s declaration in support of the habeas corpus petition stated that the computerized technology “did not exist either in the forensic literature [or] forensic casework at the time of the 1997 trial of Mr. Richards,” and that “the methods [he and Dr. Johansen used] were peer reviewed and published after [petitioner’s] conviction.”
The dissent argues that by using the word “definitively” in this context, we are imposing too high a burden of proof on petitioner, and that petitioner need only prove the falsity of Dr. Sperber’s trial testimony by a preponderance of the evidence. (Dis. opn., post, at pp. 976-979.) The dissent discusses by way of comparison this court’s decision in In re Malone (1996) 12 Cal.4th 935 [50 Cal.Rptr.2d 281, 911 P.2d 468], in which a prosecution informant made conflicting statements regarding whether he had fabricated his trial testimony. Noting that the informant had “a practice of fabricating information on pending criminal matters” (id. at p. 963), Malone accepted the referee’s finding that the petitioner had proved the falsity of the informant’s trial testimony by a preponderance of the evidence (id. at pp. 962-963).
We agree with the dissent that the preponderance of the evidence standard applies. But the dissent fails to appreciate the difference between eyewitness testimony from a witness who had a practice of fabrication and good faith expert opinion about a question as elusive as what may have caused an indistinct bruise, which is the matter at issue here. One does not prove a subjective opinion false by presenting testimony conceding its possible truth. That is particularly so when, as here, the opinion being proved false was highly tentative at the outset (asserting that petitioner’s dentition is “consistent with” the bite mark) and the opinions being used to prove its falsity are equally tentative (asserting, for example, that the expert “would tend to rule out [petitioner]”). It is as if a meteorologist’s forecast that “there could be some rain today” were being disproved by another meteorologist’s forecast that despite the heavy clouds, “I tend to think it won’t rain.” Whatever one might think of the second meteorologist’s forecast, it does not prove the first forecast false.
Thus, the dissent confuses the standard of proof with the question of what must be proved. The falsity of the trial evidence must be proved, and in the case of a tentative opinion regarding a subjective question, the opinion is not proved false if, as here, the petitioner’s experts concede it might be true. Otherwise, every criminal case becomes a never-ending battle of experts over subjective assertions that can never be conclusively determined one way or the other.