EX PARTE NAPPER

Third, the evidence was not of "such a nature that the defendant would have been unable to obtain comparable evidence by other reasonably available means." There were a limited number of ways in which the machine could malfunction — faulty calibration, extraneous interference with machine measurements, and operator error — and the defendants could raise these issues on cross-examination after examining the machine and the weekly calibration results.

See 467 U.S. at 487-90, 104 S.Ct. 2528.

373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963).

Trombetta, 467 U.S. at 488, 104 S.Ct. 2528.

Id.

Id. at 489, 104 S.Ct. 2528.

Id.

Id.

Id. at 490, 104 S.Ct. 2528.

After Youngblood the Supreme Court decided Fisher, which involved a complaint about the destruction of a substance alleged to be cocaine. Four tests by a police crime lab confirmed that the substance was cocaine. The Court held that the substance "was plainly the sort of `potentially useful evidence' referred to in Youngblood, not the material exculpatory evidence addressed in Brady" because "[a]t most, respondent could hope that, had the evidence been preserved, a fifth test conducted on the substance would have exonerated him." The defendant did not show bad faith because "police testing indicated that the chemical makeup of the substance inculpated, not exculpated, respondent" and "it is undisputed that police acted in `good faith and in accord with their normal practice.'" Reversing an Illinois appellate court, the Supreme Court held that the existence of a pending discovery request does not eliminate the necessity of showing bad faith.

Fisher. 540 U.S. at 545, 124 S.Ct. 1200.

Id.

Id. at 548, 124 S.Ct. 1200 (emphasis in original).

Id.

id.

b. "Potentially Useful Evidence"

Applicant's reliance on Fisher for the proposition that the term "potentially useful evidence" is broad is somewhat misplaced. None of the Supreme Court's cases, including Fisher, purport to provide a stand-alone definition of the term "potentially useful evidence." Rather, the Court's cases appear to contrast evidence that is at best "potentially useful" with "material, exculpatory" evidence under Brady. When evidence is at best "potentially useful," then the defendant must at least show "bad faith." That does not really answer the question of when the potential exculpatory value of evidence is so attenuated that even a showing of bad faith will not afford a basis for relief.

Professors Dix and Dawson suggest that the "possibility" of exculpation is not enough — that the defendant "must apparently show at least a substantial or considerable likelihood the evidence, if preserved, would have tended to show his innocence." A few isolated cases from other jurisdictions have held that evidence is not even potentially useful when the potential exculpatory value of the evidence is "mere speculation," when the allegedly exculpatory fact was legally irrelevant to the crime charged, when the allegedly exculpatory fact would not really provide an exculpatory inference, when the exculpatory relevance of the evidence was not shown, when the chance that the evidence would exonerate the defendant was "virtually nil," or when other evidence conclusively established that testing would not yield an exculpatory result.

Dix and Dawson, § 22.64.

United States v. Jobson, 102 F.3d 214, 219 (6th Cir. 1996) (no evidence that destroyed dispatch tape had contained any exculpatory information); Grady v. State, 2008 WY 144, P33, 197 P.3d 722, 732-33 (2008).

United States v. Williams, 577 F.3d 878, 882-83 (8th Cir. 2009) (firearm did not have to be operational to support conviction under firearm charge, so evidence that firearm may not have been operational was not potentially exculpatory).

United States v. Smith, 534 F.3d 1211, 1224 (10th Cir. 2008) (serial numbers would not have proven defendant's innocence but shown only "that perhaps she handled her money differently than a typical drug dealer"); Torres v. Mullin, 317 F.3d 1145, 1161 (10th Cir. 2003) (proving fingerprint did not belong to defendant had no exculpatory value because fingerprint could have been made by a number of people who lived in or visited the home).

Williams v. United States, 881 A.2d 557, 565 (D.C.App. 2005) (defendant did not demonstrate how revolver owned by drug rival would have been exculpatory).

Wenzel v. State, 306 Ark. 527, 533, 815 S.W.2d 938, 941 (1991) (DNA testing by the FBI of vaginal swabs containing semen).

State v. Dulaney, 493 N.W.2d 787, 791 (Iowa 1992) (plasma sample showed alcohol concentration consistent with blood sample, so no valid claim that further testing of blood sample could yield exculpatory results).

c. "Bad Faith"

More effort has been expended trying to determine the meaning of "bad faith." As Dix and Dawson acknowledge, "Precisely what constitutes `bad faith' is not clear." The suggestion by the habeas court in this case that the test for bad faith is whether the officials "were actually aware their action . . . would result in the loss of . . . evidence" miscomprehends Professors Dix and Dawson's thoughtful discussion. Dix and Dawson state that a defendant "must apparently prove at least that law enforcement officials were actually aware their action or inaction would result in the loss of what they recognized would be evidence." They do not say that such awareness is necessarily enough. If awareness that evidence would be destroyed were enough, by itself, to constitute bad faith, then officials could never conduct a test that would consume all of the evidence, even if doing so were necessary to achieve probative results. Dix and Dawson suggest that Trombetta articulates at least two situations in which bad faith would be shown: when the prosecution acts with "official animus" toward the defendants or acts with "a conscious effort to suppress exculpatory evidence."

Dix and Dawson, § 22.63.

Id.

Id.

A review of federal circuit courts and state courts of last resort in other jurisdictions reveals that they generally require a showing along those two lines. Some courts have expressly adopted the Trombetta formulation as the definitive test of bad faith. Other courts have focused on one or both of the Trombetta situations, characterizing bad faith as involving improper motivation or malice, or defining bad faith as an intent to deprive the defendant of access to (exculpatory or potentially exculpatory) evidence. Various jurisdictions have also indicated that bad faith is not established by a mere showing that the government agent was grossly negligent, engaged in intentional conduct, did not follow proper procedures, exercised poor judgment, or performed sloppy work. Bad faith was held to beabsent when government agents did not suspect that the evidence they destroyed was or might have been exculpatory, when destruction of the evidence was due to lack of training or ignorance of a recently passed law, when the government agent believed his tactics were lawful, or when the evidence was destroyed before the government agent was notified that it might have exculpatory value, before the defendant was a suspect, or before the defendant expressed a desire to have the evidence preserved. Knowing the evidence is obviously relevant to, or even determinative of, guilt or innocence has been held to be insufficient to show bad faith. The government's ability to preserve the evidence has also been held not to be dispositive.

Jobson, 102 F.3d at 218; United States v. Chaparro-Alcantara, 226 F.3d 616, 624 (7th Cir. 2000); Stuart v. State, 127 Idaho 806, 816, 907 P.2d 783, 793 (1995); State v. Lindsey, 543 So.2d 886, 891 (La. 1989).

United States v. Garza, 435 F.3d 73, 75 (1st Cir. 2006) (Defendant must show "independent evidence that the [government] was somehow improperly motivated."); United States v. Thompson, 130 F.3d 676, 686 n. 17 (5th Cir. 1997) (no demonstration of malice in tape's destruction); Gausvik v. Perez, 345 F.3d 813, 818 (9th Cir. 2003) (The defendant must "put forward specific, nonconclusory factual allegations that establish improper motive."); Collins v. Commonwealth, 951 S.W.2d 569, 573 (Ky. 1997) (Defendant "cannot substantiate any ill motive or intention on the part of the Commonwealth in failing to collect the towel."); Murray v. State, 849 So.2d 1281, 1286 (Miss. 2003) (quoting Black's Law Dictionary: "bad faith" means the "conscious doing of a wrong because of dishonest purpose or moral obliquity; it is different from the negative idea of negligence in that it contemplates a state of mind affirmatively operating with furtive design or ill will").

Featherstone v. Estelle, 948 F.2d 1497, 1505 (9th Cir. 1991) (finding that destruction of lineup photo was not in bad faith because "it was not deliberately done to deprive petitioner of access to relevant evidence"); People v. Seaton, 26 Cal.4th 598, 656-57, 110 Cal. Rptr.2d 441, 28 P.3d 175. 209 (2001) (no showing that police believed the unpreserved blood evidence would have exculpated defendant or "that their purpose was to deny him the opportunity to use the evidence to exculpate himself"); United States v. Day, 697 A.2d 31, 35-36 (D.C.App. 1997) ("Nothing in the record on appeal hints, even remotely, that the police destroyed the car so that [defendant] could not examine it, or have it viewed by an expert."); Dufour v. State, 905 So.2d 42, 68 (Fla. 2005) (quoting Guzman v. State, 868 So.2d 498, 509 (Fla. 2003): "Under Youngblood, bad faith exists only when police intentionally destroy evidence they believe would exonerate a defendant."); Guzman, 868 So.2d at 510 (no showing that "any State actor intentionally deprived [defendant] of evidence which the Stale actor believed to be exculpatory"); Dulaney, 493 N.W.2d at 791 (no evidence State "intentionally destroyed" blood sample "in an effort to deprive [defendant] of evidence"); State v. Berkley, 567 A.2d 915, 918 n. 3 (Me. 1989) (collusion designed to prevent independent testing would establish bad faith); Slate v. Bailey, 677 N.W.2d 380, 393 (Minn. 2004) ("no suggestion that the State destroyed or released items `to avoid discovery of evidence beneficial to the defense'"); Murray, 849 So.2d at 1286 (bad faith where "destruction was intentional and indicates fraud and a desire to suppress the truth"); State v. Hernandez, 2005 ND 214, P31, 707 N.W.2d 449, 461 (2005) (defining "bad faith" to mean "the evidence was deliberately destroyed by or at the direction of a State agent who intended to thwart and to deprive the defense of information."); State v. Bousum, 2003 SD 58, P16, 663 N.W.2d 257, 263 (2003) (adopting North Dakota's definition of "bad faith"); Drury v. State, 2008 WY 130, P19, 194 P.3d 1017, 1023 (2008) (In determining bad faith, "we must take into account the State's knowledge of the potential exculpatory nature of the evidence.").
The Supreme Court of North Dakota has indicated that, although not constituting "bad faith," if there were shown to be a "systemic disregard" of the "State's duty to zealously protect evidence in its possession" so that "haphazard handling and destruction of evidence . . . is commonplace," then "prophylactic measures, such as an adverse-inference instruction" may be justified. City of Bismarck v. Holden, 522 N.W.2d 471, 475 (N.D. 1994) (ellipsis in original).

Garza, 435 F.3d at 75 (citing United States v. Femia, 9 F.3d 990, 995 (1st Cir. 1993)); United States v. Branch, 537 F.3d 582, 590 (6th Cir. 2008).

Garza, 435 F.3d at 75 ("While the evidence was destroyed as a result of Quinn's conscious and deliberate decision, intentionality is not enough to show bad faith."); Bailey, 677 N.W.2d at 393 ("destruction of the evidence was `intentional'" but was not in bad faith).

Garza, 435 F.3d at 76 (no bad faith even though "it was not the `normal' procedure to destroy evidence in cases that were still open"); United States v. Deaner, 1 F.3d 192, 200 (3rd Cir. 1993) ("[W]e have not held that an improper procedure in and of itself implies bad faith. While it may permit such an inference, it does not syllogistically imply the presence of bad faith as a matter of deductive logic."); Guzman, 868 So.2d at 509 ("bad faith does not turn on whether law enforcement officers followed established procedures"); State v. LaMae, 268 Kan. 544, 551, 998 P.2d 106, 111 (2000) (failure of the DEA to follow its guidelines, though troubling, not enough to show bad faith).

Lovitt v. True, 403 F.3d 171, 187 (4th Cir. 2005) (court clerk exercised "serious error in judgment" in destroying evidence); Murray, 849 So.2d at 1286 (bad faith is "not simply bad judgment or negligence"); State v. Hanna, 123 Wash.2d 704, 714, 871 P.2d 135, 140 (1994) (failure to preserve and retain evidence may have reflected "poor judgment" but was based on a reasonable belief that the evidence would not be relevant to either party).

United States v. Carreno, 363 F.3d 883, 888 (9th Cir. 2004) (Sloppy work is not tantamount to bad faith.).

United States v. Stevens, 935 F.2d 1380, 1388 (3rd Cir. 1991) (test that destroyed the evidence "might have inculpated or exculpated" the defendant but "no one knew," thus nothing suggested government agents "suspected that the saliva/semen sample . . . could form a basis for exonerating" the defendant); Jobson, 102 F.3d at 218 ("no evidence that anyone in the Detroit Police Department or the U.S. Attorney's office suspected that the tape was exculpatory"); Carreno, 363 F.3d at 888 ("no evidence government knew, or even suspected, that" deported witness "would testify in [the defendant's] favor").

United States v. Ramos, 27 F.3d 65, 71-72 (3rd Cir. 1994) (due to inadequate training, government agent affiliated with the DEA believed he was following office procedure in destroying notes); United States v. Houston, 548 F.3d 1151, 1155 (8th Cir. 2008) (officers not trained to download video evidence).

Lovitt, 403 F.3d at 187 n. 3; Lovitt v. Warden, 266 Va. 216, 242, 585 S.E.2d 801, 816 (2003).

Gausvik, 345 F.3d at 818.

United States v. Beckstead, 500 F.3d 1154, 1159-60 (10th Cir. 2007) (Government destroyed the evidence before the defendant had opportunity to notify the officials that the alleged drug lab and its chemicals were potentially exculpatory and it was unlikely that further testing of the lab would produce exculpatory results); United States v. Bohl, 25 F.3d 904, 910 (10th Cir. 1994) (discussing United States v. Richard, 969 F.2d 849 (10th Cir. 1992) where government destroyed boxes of marijuana and the defendant did not show that the government knew he wanted them preserved for trial); Sharp v. State, 286 Ga. 799, 802, 692 S.E.2d 325, 329 (2010) (State unaware of exculpatory value of condom that was lost "before [defendant] was arrested or identified as a suspect"); State v. Eason, 133 N.H. 335, 342, 577 A.2d 1203, 1207 (1990) (when police destroyed what they thought was an animal carcass, they did not know victim had been murdered, much less that the defendant murdered him).

Monzo v. Edwards, 281 F.3d 568, 580 (6th Cir. 2002) ("It is not enough that the police knew that semen samples could be determinative of guilt or innocence if preserved or tested."); State v. Osakalumi, 194 W.Va. 758, 765, 461 S.E.2d 504, 511 (1995) (even though police acted negligently in disposing of a couch [with a bullet hole] that "was so obviously a part of a pending police investigation," court could not say their actions were motivated by bad faith).

Beckstead, 500 F.3d at 1161.

But courts have found bad faith where government agents hid or concealed the evidence until it was destroyed, destroyed the evidence after being expressly notified by the defense of its potential exculpatory value, destroyed the evidence after a dismissal of the case (and then reinstituted proceedings), or where government agents refused to show up in court for an inquiry into their good or bad faith. The Tenth Circuit has formulated a five-factor test that considers whether: (1) the government received explicit notice of the evidence's potentially exculpatory nature, (2) the claim was backed up with objective, independent evidence rather than being conclusory, (3) the evidence was in the government's control at the time of notice, (4) the evidence was central to the case, and (5) there is any innocent explanation for the evidence's destruction.

Yarris v. County of Delaware, 465 F.3d 129, 133, 142-43 (3rd Cir. 2006) (plaintiff's petition alleged that detectives refused a request to deliver slides to coroner but kept them in a paper bag under a desk until the biological material rotted and became useless for DNA testing); Stuart, 127 Idaho at 816, 907 P.2d at 793 ("There is a clear recognition that concealment is one method of proving that the exculpatory value of the evidence was known to the government prior to its destruction." Prosecution concealed the existence of a tape recording of a phone call that would have led to phone logs that were later destroyed.).

United States v. Cooper, 983 F.2d 928, 931 (9th Cir. 1993) ("The equipment's value as potentially exculpatory evidence was repeatedly suggested to government agents." Parole officer and alleged drug lab's landlord reported lab's claims of legitimacy during preseizure investigation. Law enforcement agents knew lab was ostensibly configured to make a legal substance rather than methamphetamine before destroying the lab's equipment.); Bohl, 25 F.3d at 911 ("[G]overnment was explicitly placed on notice that" defendants "believed the tower legs were potentially exculpatory" before government destroyed the evidence. And assertion of potentially exculpatory value was not merely conclusory but was backed up by objective, independent evidence.).

State v. Jackson, 302 S.C. 313, 316, 396 S.E.2d 101, 102 (1990).

State v. McGrone, 798 So.2d 519, 523 (Miss. 2001) (defense counsel subpoenaed police officers for two separate hearings, but the officers never appeared).

Smith, 534 F.3d at 1224-25.

Several courts have explained Young-blood's "bad faith" requirement as substituting for Brady's requirement that evidence be exculpatory. Essentially, an inference that the evidence was probably exculpatory can be drawn from a government agent's bad faith conduct.

State v. Youngblood, 173 Ariz. 502, 506, 844 P.2d 1152, 1156 (1993); State v. Craig, 490 N.W.2d 795, 797 (Iowa 1992). See also Wenzel, 306 Ark. at 533, 815 S.W.2d at 941 (quoting Yoimgblood's statement that its standard was directed to "those cases in which the police themselves by their conduct indicate that the evidence could form a basis for exonerating the defendant").

See authorities cited in previous footnote.

4. The Present Case

a. Potential Usefulness of the Evidence

The habeas court and the parties have arrived at some erroneous conclusions because they have treated the DNA evidence as a monolithic whole, instead of considering each of the four DNA samples separately. The habeas court made this mistake when it interpreted Orchid Cellmark's reference to an "unknown individual" as an indication that the link between applicant and the DNA evidence was weak. Although ReliaGene was able to obtain results from only one of the epithelial samples, Orchid Cellmark was able to obtain results from both epithelial samples. For both samples, Orchid Cellmark detected a primary DNA profile that matched the victim. One of the samples also contained some minor alleles matching applicant's profile. The other epithelial sample contained minor alleles that did not match either the victim or applicant. These alleles belonged to another individual (or individuals), whose identity was, of course, "unknown." Subsequent testing was done to attempt to rule in or rule out individuals who were involved in DNA testing as a possible source of contamination. Notably, Orchid Cellmark was able to rule out the persons from HPD who were originally involved in collecting and testing the material: Verbitskey, Hilleman, Childs-Henry, and Chu did not contribute the "unknown person's" DNA.

Although an unknown person's DNA was detected in one of the epithelial samples, that fact does not have any tendency to exculpate applicant. Because the epithelial samples are composed of cells that are easily shed, such as skin cells, the DNA in question did not necessarily come from the perpetrator — it could have come from anyone who had recent contact with the victim. It is also possible that the extraneous alleles are the result of contamination, or the "allele drop-in" (referred to by Pineda) that might occur as a result of repeated testing. Had the sperm samples yielded an "unknown person's" DNA, that fact might have been significant, but that is not what happened.

The same fallacy of treating the DNA results as a monolithic whole also underlies the State's and applicant's competing positions regarding why the outside labs' test results differ from those obtained by HPD. Although the State contends that the difference in results is due to the quantitative decline of the sample, applicant questions how that can be, when the outside labs were able to obtain a complete profile matching the victim. The answer to applicant's question is simple once the samples are treated separately: The epithelial samples have consistently matched the victim throughout all of the testing but have never provided strong results with respect to the perpetrator. Chu's STR testing detected a single allele in one of the epithelial samples that was "unique" to applicant's profile (i.e. not possessed by the victim). Chu's testing of the other epithelial sample revealed no unique alleles. The one unique allele that was detected was one of nine unique alleles possessed by applicant that was subject to testing.

The outside labs actually detected a greater number of alleles unique to applicant's profile. ReliaGene detected two of twelve tested-for unique alleles, while Orchid Cellmark detected four of fourteen tested-for unique alleles. Although the habeas record does not contain a statistical calculation for the random match probability of the single unique epithelial allele detected by Chu, it seems likely that the four of fourteen unique alleles detected by Orchid Cellmark yields a stronger link to applicant than the one of nine unique alleles detected in HPD's testing.

Even if that one unique allele detected by Chu somehow carried special statistical significance that provided a stronger link to applicant than the four unique alleles detected by Orchid Cellmark (and the two detected by ReliaGene), that would not make the Orchid Cellmark (or ReliaGene) results exculpatory. ReliaGene's and Orchid Cellmark's failure to detect the unique allele detected in Chu's testing is easily explained, as the State points out, by the quantitative decline of the sample. Although Chu had access to a liquid extract for his testing, ReliaGene and Orchid Cellmark were forced to use a buffer solution to resuspend possible trace amounts of DNA attached to the walls of seemingly empty tubes.

Because the epithelial DNA was originally derived from cells from the skin of the victim's face, one would expect the victim's cells in the epithelial sample to be, by far, the most abundant, explaining the outside labs' ability to obtain a full DNA profile matching the victim, even from the trace amounts of DNA. Chu's ability to detect only one foreign allele suggests that the epithelial samples contained only a small amount of DNA extraneous to the victim. So it is not surprising that this single foreign allele would be lost after the massive decline in the quantity of the sample. Likewise, Orchid Cellmark's failure to detect one of the alleles detected by ReliaGene could have resulted from the further decline in the sample caused by the second round of tests.

If the testing of the three labs is considered together, a stronger link to applicant is established. Each lab detected at least one unique allele that was not detected by the others. Such results could be explained by the loss of some alleles caused by a quantitative decline in the sample, combined with the detection of additional alleles through progressively more sensitive techniques (or perhaps by chance). If all of the detected alleles that correspond to the tested loci are considered without regard to which lab detected them, then there were six alleles unique to applicant out of fourteen that he possessed that were subjected to tests. We do not know if a statistical calculation could be done that incorporated all three labs' results. We simply point out that the discovery of diverse alleles unique to applicant by the three labs tends to inculpate applicant. So, the further testing of the epithelial samples seems to have produced a stronger inculpatory inference than the epithelial samples originally provided.

It is nevertheless true that the epithelial results do not approach the strength of Chu's results for the sperm samples. The outside labs were not able to obtain any results for the sperm samples. The failure to obtain results is not exculpatory in itself; it might simply mean that the samples had degraded to the point where retesting could not produce any results. But with respect to the strength of the test results, the parties and the habeas court have failed to appreciate the differences between the sperm samples and the epithelial samples. It is true that ReliaGene and Orchid Cellmark produced results that are much weaker statistically than some of the results obtained by Chu. Applicant sees this difference in the strength of the results as indicating that the DNA evidence shows the link to applicant to be less substantial than previously thought, while the State chalks up this difference to the loss of alleles caused by a quantitative decline in the sample. Neither position is correct. The difference in the statistical strength of the results is a consequence of the fact that different samples are at issue. Chu relied upon the sperm samples at trial, and Bromwich's statistical calculation of 1 in 232,000 was undoubtedly based upon the sperm samples. The results obtained by the outside labs were for the inferior epithelial samples. For the epithelial samples, it appears that the outside labs obtained stronger results than Chu obtained, and discovered additional alleles possessed by applicant that were not previously discovered in the HPD testing. Although we do not entirely agree with the State's reasoning, we agree that the subsequent test results tend to confirm applicant's guilt.

In assessing whether the consumed DNA material constituted "potentially useful" evidence, we consider the likelihood that a defense expert would have produced exculpatory results if given the opportunity to test sperm and epithelial portions of the original liquid DNA extract. We conclude that the likelihood of an exculpatory result is small, though not zero. Although Bromwich criticized Childs-Henry's lack of documentation and would have preferred that she run more definitive tests for semen, he did not suggest that she obtained an incorrect result, or that any flaw existed in her analysis that would have undermined the validity of Chu's results. As we have explained, Bromwich found that Chu had "developed clean, interpretable profiles" that were "good quality and clear results from potentially difficult forensic evidence samples." Bromwich calculated a 1 in 232,000 random match probability for the African-American population that was "relatively strong" given the nature of the sample. We also note that it is virtually impossible for applicant's DNA to have contaminated the original test results because the swab DNA was typed before applicant was even a suspect.

Nevertheless, in a city the size of Houston, a 1 in 232,000 probability raises the possibility that a handful of people might perhaps be included in the DNA results. A defendant in applicant's position would hope that testing by a defense expert would detect additional alleles inconsistent with the defendant's DNA profile. And given the now-tarnished reputation of the HPD Crime Lab, a defendant might even hope that a defense expert would obtain dramatically different results, revealing some unknown malfeasance by the lab.

The retesting by ReliaGene and Orchid Cellmark tend to undercut any such hopes, as the retesting has produced results consistent with applicant's DNA. But, while contamination of the original HPD test results with applicant's DNA is a virtual impossibility, it is possible that contamination could have occurred later, in time to skew the retesting results. Dr. Johnson believed that to be a possibility, Orchid Cellmark's "unknown person" result for one of the epithelial samples may be an example of contamination, and the HPD Crime Lab's act of forwarding an incorrect reference sample to ReliaGene hardly inspires confidence. The possibility of contamination may also be increased somewhat by the fact that the tubes were seemingly empty, with relevant Crime Lab personnel believing that the evidence had been consumed, so the HPD Crime Lab may not have had as much incentive to maintain the integrity of the evidence.

Given the paucity of authority and the brevity of the pronouncements regarding whether evidence is even "potentially useful," we are not certain whether an inquiry into potential usefulness would include looking at other evidence at trial tending to show that the defendant was guilty and, therefore, that DNA testing by a defense expert would not likely yield exculpatory results. The background portion of this opinion outlines the substantial, persuasive evidence of applicant's guilt aside from the DNA test results, and we will discuss this evidence in connection with the ineffective assistance of counsel claim later in this opinion. This inculpatory evidence would further support a conclusion that the likelihood of exculpatory results from defense DNA testing was slim.

Nevertheless, we refrain from deciding whether the consumed DNA evidence in this case meets a threshold showing of potential usefulness. That the likelihood of exculpatory results was slim is a factor that may be taken into account when assessing the existence or absence of bad faith. Where exculpatory results were unlikely, an inference can be drawn that the DNA analyst was probably not acting with the intent to deprive the defendant of exculpatory evidence when he destroyed the sample. A threshold showing of potential usefulness may be a worthy concept to apply in a future case; we simply decide here to move on to the question of bad faith.

b. "Bad Faith"

The habeas court's finding of "bad faith" is based in part upon what we have already recognized to be an inaccurate definition. "Bad faith" is more than simply being aware that one's action or inaction could result in the loss of something that is recognized to be evidence. As the cases we have discussed show, bad faith entails some sort of improper motive, such as personal animus against the defendant or a desire to prevent the defendant from obtaining evidence that might be useful. Bad faith cannot be established by showing simply that the analyst destroyed the evidence without thought, or did so because that was the common practice, or did so because the analyst believed unreasonably that he was following the proper procedure.

No evidence in this case suggests that Childs-Henry or Chu acted in bad faith. Chu specifically stated in his affidavit that his consumption of the evidence was not the result of a desire to prevent additional testing. The habeas court was not required to believe that affidavit, but to find bad faith, a finder of fact must do more than simply disbelieve proffered evidence of good faith: There must be some evidence from which an inference of bad faith can be drawn. There is no such evidence here. Neither analyst has been shown to have any animus against applicant — applicant was not even a suspect at the time testing was conducted. Nor is there any evidence that the analysts were aware of any impropriety in their own conduct, such as a history of falsifying test results. The Bromwich Report explained that the failures in the HPD Crime Lab were the result of a lack of training and of the isolation of the DNA section of the Crime Lab from the forensic community, not the actions of "rogue analysts."

Cf. Boyle v. Johnson, 93 F.3d 180, 185-87 (5th Cir. 1996) (referring to accusations that Ralph Erdmann had falsified autopsies and committed perjury in prior cases); State v. Clifford, 328 Mont. 300, 321 n. 4, 121 P.3d 489, 503 n. 4 (2005) (referring to Erdmann and Fred Zain as examples of experts who misstated, altered, or falsified results).

The habeas court gave three reasons for finding bad faith. First, it concluded that the analysts must have been aware of a law that was enacted but not yet effective. But the history of Article 38.39 reveals that the statute was not enacted until after the HPD Crime Lab's DNA testing in this case. The legislation was originally introduced on January 17, 2001. The bill passed the Senate on February 19, 2001, and passed the House in an amended version on March 22, 2001. After the bill was sent to a conference committee, it passed both houses of the Legislature on April 3, 2001. The bill was signed by the Governor on April 5, 2001, and made effective immediately. As we explained earlier, the DNA from the swab samples was typed on February 20, 2001. We find no basis for the habeas court's conclusion that the analysts were aware of this impending legislation.

Texas Legislature Online History, Bill: SB 3, Legislative Session: 77(R), Council Document: 77R 1805 GWK-F.

Id.

Id.

Id.

Regarding the FBI and NRC recommended standards, we agree that the HPD Crime Lab's analysts should have been aware of them. But given the lack of training and the isolation of the crime lab from the forensic community, it is questionable whether the analysts were actually aware of them. Even if they were, there is no evidence to suggest that they believed that the standards regarding preservation of a sample would apply in this case.

When Childs-Henry discarded the original swabs, there were redundant samples of DNA. Although it would have been better to retain the original swabs, the existence of redundant samples would appear to satisfy the FBI and NRC guidelines.

The evidence is clear that, unlike Childs-Henry, Chu understood that his work would consume all of the remaining sample. The habeas court cited Chu's explanation that he had to consume the sample due to the small amount of DNA, but Chu further testified at trial to conducting a cross analysis of the swabs, which could not have occurred unless he tested the material from both swabs, which in turn would require consuming all of the sample. As the Bromwich Report explains, Chu made a mistake in testing samples from both swabs. He should have saved one of the epithelial samples and one of the sperm samples for later testing, which could have been conducted by another DNA lab. But the Bromwich Report does not suggest that Chu's failure to do so was anything other than a mistake.

Finally, though we accept that the habeas court can, from its own recollection, recount common practices within that court, we cannot accept the habeas court's conclusion that the HPD Crime Lab analysts must have been aware of a practice in Harris County regarding any notification that evidence would be consumed in testing. As an example of this practice, the habeas court cited the signed acknowledgment arising in the instant habeas proceedings regarding Orchid Cellmark's anticipated testing. This document, executed years after the public disclosure of serious problems with the HPD Crime Lab, is not particularly probative of the practice followed before these problems came to light. Further, given the trace amounts of DNA and ReliaGene's use of a buffer solution, ReliaGene's testing could have destroyed any remaining sample. At the very least, Pineda's affidavit suggests that ReliaGene's testing could have been expected to degrade the sample further, resulting in a further loss of alleles for any subsequent testing. Yet, there was no acknowledgment regarding ReliaGene's anticipated testing.

Moreover, we see a practical problem with the idea that analysts should have notified the prosecutor of the possible consumption of the evidence in this case. HPD's DNA testing was conducted, and the liquid extract was consumed, before there were any suspects. Whatever the practice of the Harris County Criminal District Courts, there was no defense attorney for the District Attorney's Office to notify at the time DNA testing was conducted. Given the lack of training received by the HPD Crime Lab's DNA analysts, nothing in the record suggests that Childs-Henry and Chu were aware of any need to notify the District Attorney's Office before consuming the sample, especially since no prosecution was pending.

B. Perjury/False Testimony Claim 1. The Parties' Arguments and the Habeas Court's Findings

Applicant claims that Chu's statistical frequency estimate constituted "false testimony" and "possible perjury" by a prosecution witness in violation of the Fourteenth Amendment to the United States Constitution. To support his claim, applicant relies upon statements in the Bromwich Report, a letter from Orchid Cellmark, and statements made by Dr. Johnson in her habeas affidavit. From the Bromwich Report, applicant cites statements that Chu's statistical frequency estimate was "incorrect and misleading" and based upon a practice that was "extremely flawed." Applicant characterizes Orchid Cellmark's statistical frequency estimate of 1 in 255 for African-Americans as "the true statistical frequency for black persons." And applicant cites Dr. Johnson's statements that Chu's trial testimony was "completely unsupported scientifically" and "patently false," as well as her statement that a mixture of DNA "cannot possibly statistically match only one person."

Finding in the State's favor on this claim, the habeas court concluded: "Because the State did not knowingly use false or perjured testimony at trial, the applicant failed to show that he was denied Due Process under the Fourteenth Amendment to the United States Constitution."

The habeas court cited Ex parte Castellano, 863 S.W.2d 476, 480 (Tex.Crim.App. 1993); Mooney v. Holohan, 294 U.S. 103, 112, 55 S.Ct. 340, 79 L.Ed. 791 (1935); and Pyle v. Kansas, 317 U.S. 213, 216, 63 S.Ct. 177, 87 L.Ed. 214 (1942).

Applicant argues in his brief that "logic suggests that . . . if the DNA found on the evidence was a mixture of more than one person's DNA, then it could not have come from only one person." Applicant then echoes statements from the Bromwich Report and from Dr. Johnson to the effect that Chu's testimony was "contradictory, incorrect, misleading, completely inappropriate . . ., internally contradictory . . ., unsupported scientifically" and "patently false." Applicant contends that Chu, an agent of the State, committed perjury and that the prosecutor unknowingly presented Chu's false testimony. Applicant also argues that the testimony was harmful because DNA is powerful evidence and Chu's scientific conclusion was dramatic and cast doubt on all of the defensive evidence as well. Applicant also points to Orchid Cellmark's 1 in 255 estimate and the Bromwich Report's 1 in 232,000 estimate as differing from Chu's claim that no one on Earth but applicant could have committed the crime.

The State counters that Chu's testimony was not perjury because no evidence shows that Chu had an intent to deceive or knew that his statistical calculations were false. Characterizing Chu's calculations as "false" or "arguably false," the State contends that applicant is required to show by a preponderance of the evidence that the testimony contributed to his conviction or punishment and that he did not do so. The State discusses a great deal of the evidence introduced at trial and concludes that "ample evidence," aside from Chu's testimony, demonstrates applicant's guilt.

2. Preservation of Error

Applicant made no complaint about the alleged falsity of Chu's testimony at trial. There is at least a question regarding whether applicant has forfeited his complaint as a result. This question is complicated by the fact that the State has a continuing "duty to correct `false' testimony whenever it comes to the State's attention." Without expressing any opinion on the preservation of error question, we will address the merits.

See Estrada v. State, 313 S.W.3d 274, 288 (Tex.Crim.App. 2010).

Id. at 288.

3. The Law

How we categorize applicant's claim is crucial to determining the standard of materiality or harm to be applied.

(a) Knowing use of perjured testimony-direct appeal

A prosecutor who procures a conviction through the knowing use of perjured testimony violates due process:

[Due process is violated] if a State has contrived a conviction through the pretense of a trial which in truth is but used as a means of depriving a defendant of liberty through a deliberate deception of court and jury by the presentation of testimony known to be perjured. Such a contrivance by a State to procure the conviction and imprisonment of a defendant is as inconsistent with the rudimentary demands of justice as is the obtaining of a like result by intimidation.

A conviction is considered to have been procured through the use of perjured testimony if the perjured testimony is material. The standard of materiality for a prosecutor's knowing use of perjured testimony is the harmless error standard articulated in Chapman v. California: the evidence is material (and harmful) unless it can be determined beyond a reasonable doubt that the testimony made no contribution to the defendant's conviction or punishment. This most favorable materiality or harm standard is available to a defendant on direct appeal.

386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967).

Ex parte Fierro, 934 S.W.2d 370, 372-73 (Tex.Crim.App. 1996).

Id. at 373.

(b) Knowing use of perjury -habeas corpus

On habeas corpus, the standard of harm that must be satisfied to obtain relief may be more burdensome to the defendant than the standard of materiality that must be satisfied to establish a due process violation. In Ex parte Fierro, we determined that, where the habeas applicant "had the opportunity to discover the basis of his claim in time to advance it at trial or in a motion for new trial," he must show "by a preponderance of the evidence that the error contributed to his conviction or punishment." In other words, we employed the normal standard on habeas review, with the burden on the applicant.

Id. at 373-74.

Id. at 374-75, 374-75 n. 10.

See id. at 372.

Where the habeas applicant lacked such an opportunity, we left open the possibility that he might "avail himself of the Chapman harmless error standard."

Id. at 375 n. 10.

(c) Unknowing use of perjured testimony" habeas corpus

We have recently held that due process may be violated by a prosecutor's unknowing use of perjured testimony. In Exparte Chabot, the applicant had no prior opportunity to discover the basis for his claim that DNA evidence proved that the State's witness had perjured himself. In discussing what harm standard to employ under those circumstances, we said, "[W]e see no reason for subjecting the two types of errors [knowing versus unknowing use of perjured testimony] to different standards of harm." That is, the preponderance of the evidence standard applies, on habeas review, to either the knowing or the unknowing use of perjured testimony.

Chabot, 300 S.W.3d at 771-72; see also Ex parte Carmona, 185 S.W.3d 492 (Tex.Crim. App. 2006).

Chabot, 300 S.W.3d at 770.

Id. at 771.

Chabot did not settle the question left open in Fierro regarding whether a more favorable standard of harm may apply to the knowing use of perjured testimony that the habeas applicant had no prior opportunity to discover. Rather, Chabot does not appear to have taken into account the caveat in Fierro, and so Chabot simply stands for the proposition that the preponderance of the evidence standard is appropriate for the unknowing use of perjured testimony that the habeas applicant had no prior opportunity to discover. It remains unsettled whether a more favorable standard might be available to a defendant in the "knowing use" context.

The Second Circuit has suggested that the unknowing use of perjured testimony violates due process if a stronger showing of materiality or harm is made than is required by the Chapman standard. Sanders v. Sullivan, 863 F.2d 218, 225-26 (2nd Cir. 1988).

(d) False testimony as opposed to perjured testimony

In Estrada v. State, we held on direct appeal that false testimony that was not perjury resulted in a due process violation when there was "a fair probability that [the] death sentence was based upon . . . incorrect testimony." In that case, the State's witness gave inaccurate testimony about the prison classification system but was not aware of the inaccuracy. The jury sent out a note regarding the very subject matter of this testimony. It was later determined that the testimony was inaccurate, and the State confessed error on direct appeal. In sustaining the defendant's contention, we held that his complaint was not forfeited because the defendant "could not reasonably be expected to have known that the testimony was false at the time that it was made."

313 S.W.3d at 287.

Id. at 286-87.

Id.

Id.

Id. at 288.

(e) False testimony in the actual innocence context

Finally, we observe that our "actual innocence" jurisprudence could encompass a claim on habeas that involved newly discovered evidence that a witness's testimony was false. Under that jurisprudence, the habeas applicant must show "by clear and convincing evidence that, presented with both the inculpatory evidence at trial and the newly discovered or available evidence of innocence, no reasonable juror would have convicted him."

See Ex parte Elizondo, 947 S.W.2d 202 (Tex.Crim.App. 1996) (recantation of trial testimony).

Ex parte Chavez, 213 S.W.3d 320, 322 (Tex.Crim.App. 2006).

4. The Present Case

To obtain the most favorable standard of materiality or harm that might possibly apply, applicant must show that the State knowingly used perjured testimony and that he did not have an opportunity to discover the perjury in time to present a claim at trial or in a motion for new trial (or perhaps on appeal). No one has contended in this case that the prosecutors were aware that there was any problem with Chu's statistical frequency estimate. Nevertheless, we treat perjured testimony as knowingly used if the witness was a member of the "prosecution team." To obtain review under the most favorable standard of materiality or harm, then, applicant must at least show: (1) that Chu was a member of the prosecution team, (2) that Chu committed perjury, and (3) that applicant did not have an opportunity to discover the perjury in time to present his claim at an earlier proceeding.

The Supreme Court has suggested that the knowing use of "false" testimony is treated the same as the knowing use of perjured testimony. United States v. Bagley, 473 U.S. 667, 678, 679 n. 8, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985) (discussing Napue v. Illinois, 360 U.S. 264, 79 S.Ct. 1173, 3 L.Ed.2d 1217 (1959)).

Fierro, 934 S.W.2d at 372 ns.2, 3.

In Texas, the crime of perjury is committed if "with intent to deceive and with knowledge of the statement's meaning" a person makes a false statement under oath or unsworn declaration. In connection with a claim regarding the knowing use of perjured testimony, the First Circuit relied upon federal law defining perjury as "the willful intent to provide false testimony, rather than as a result of confusion, mistake, or faulty memory." The Fifth Circuit, however, has eschewed "the strict legal definition of perjury" and found "perjury" within the meaning of the prohibition against its knowing use when witnesses "did not candidly respond to the defense counsel's questions."

TEX. PENAL CODE § 37.02(a).

United States v. Tavares, 93 F.3d 10, 14 (1st Cir. 1996).

United States v. Carter, 566 F.2d 1265, 1270 (5th Cir. 1978).

But regardless of the precise contours of the meaning of "perjury" for the purpose of determining a due process violation, applicant has not demonstrated that Chu committed perjury in this case. The habeas court found against applicant on this claim, so we review the evidence deferentially in the State's favor. Nothing in the record before us suggests that Chu intended to provide false testimony or to deceive, or that he was less than candid in his testimony. In other words, we have no reason to believe that Chu thought his testimony was inaccurate. The Bromwich Report indicates that Chu's method of calculating the statistical frequency estimate, though extremely flawed, was the HPD Crime Lab's usual practice. The Bromwich Report did not suggest that HPD analysts were committing perjury every time they testified to their usual calculations in court. As we have pointed out, the report said that the problems with the HPD Crime Lab were due to poor training and not to rogue analysts. There is ample support for a finding that Chu did not commit perjury, consistent with the habeas court's conclusion that the State did not knowingly use perjured testimony. Because Chu did not commit perjury, applicant's claim does not fall within Fierro's caveat regarding knowing-use-of-perjury claims that could not have been presented previously. We again leave open that question.

It is at least questionable whether Chu was a member of the proseeution team and whether applicant lacked the opportunity to discover the perjury at trial or in motion for new trial proceedings. Given our conclusion that there was no perjury, we need not address these issues.

For habeas corpus proceedings involving the unknowing use of perjured testimony, the harm standard would be preponderance of the evidence, as we have stated above. Applicant would not be entitled to a more favorable standard for the unknowing use of false (but not perjured) testimony. The preponderance of the evidence standard is less favorable to applicant than the standard employed in analyzing an ineffective assistance of counsel claim under Strickland v. Washington As explained below, applicant fails to meet the prejudice prong of Strickland. This means that applicant's false evidence claim must also necessarily fail.

466 U.S. 668, 693, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984) ("[A] defendant need not show that counsel's deficient conduct more likely than not altered the outcome in the case. . . . The result of a proceeding can be rendered unreliable, and hence the proceeding itself unfair, even if the errors of counsel cannot be shown by a preponderance of the evidence to have determined the outcome.").

C. Ineffective Assistance of Counsel 1. The Parties' Arguments and the Habeas Court's findings

In his application, applicant contends that defense counsel Greenlee performed deficiently in failing to hire a forensic or DNA expert. Applicant further claims that, had defense counsel hired an expert: (1) Chu's statistical analysis would have been revealed at trial as false, (2) the jury would have been made aware that Chu violated the standard of care in his profession when he consumed all of the DNA evidence, and (3) defense counsel would have been able to obtain the same type of testing that was later conducted by ReliaGene and Orchid Cellmark. Applicant also contends that Greenlee performed deficiently by failing to object to the DNA test results on the basis of Chu's consumption of the sample, by failing to move for a continuance after Childs-Henry testified that there was liquid extract sample remaining so that DNA testing could then occur, and by failing to adequately cross-examine Childs-Henry and Chu at trial.

In his brief, applicant also points to Greenlee's lack of prior experience in DNA matters and suggests that he could not possibly conduct a proper investigation without the help of an expert. He also argues that there was no good reason for Greenlee to forgo an expert witness who could have enhanced the defense trial strategy of emphasizing the unfairness of the lab's consumption of all of the DNA evidence. Applicant also contends that there were many areas of possible impeachment of the State's experts that were not raised at trial. Applicant also describes as "inexplicable" Greenlee's failure to move for a continuance when ChildsHenry indicated that DNA extract still existed.

Regarding the prejudice prong of Strickland, applicant contends that, had counsel performed properly, the following information would have been learned and used at tral: (1) that Chu's statistical frequency estimate was illogical and false, (2) that the HPD Crime Lab's consumption of the entire sample was unnecessary and improper, (3) that the quality of Childs-Henry's testing for semen was poor, and (4) that obtaining DNA profiles from a mixture was fraught with difficulty. Applicant contends that this information would have changed "the entire complexion of the case."

The State counters that Greenlee made strategic decisions with plausible bases after a thorough investigation. The State points to various acts by defense counsel and his articulated strategies. This discussion by the State essentially rehashes the habeas court's findings.

The court found that Greenlee's testimony at the evidentiary hearing was credible. The court found that Greenlee had a good working relationship with the prosecutors, who provided him with whatever information he requested, and had free access to the State's files and evidence. It also found that Greenlee had no reason to doubt Glaeser's representation that the DNA evidence had been consumed, with nothing left for an independent analysis. The court found that Greenlee's decision to proceed without an expert was based on a strategy of arguing to the jury the unfairness of being denied an opportunity to conduct independent testing. The court found that this strategy was formulated after a thorough investigation that included consulting at least one other defense attorney.

The court also pointed to Greenlee's strategy of cross-examining the crime scene officer, Verbitskey, to show that he was not qualified to collect the swabs from the victim's cheek and may have exposed the evidence to contamination. The court found that defense counsel's cross-examination of Childs-Henry was strategically designed to show that she was results-oriented, could not describe the protocol and procedures she followed in performing the DNA extraction, and could not confirm that she followed the proper steps in the proper order during the procedure. The court also found that Greenlee's cross-examination of Childs-Henry elicited what Greenlee perceived to be the beneficial testimony that only a "sloppy chemist" would consume all of the evidence in performing a DNA analysis. And the court found that defense counsel cross-examined Chu with a strategy toward "demonstrating that Chu had consumed all of the DNA evidence without regard for the fact that the defense was denied an opportunity to conduct independent DNA testing." The court further found that Greenlee believed that his cross-examination succeeded in this purpose and that he did not need a consulting expert to accomplish that end.

The court also found that defense counsel emphasized in closing argument that the defense was unfairly denied the opportunity to conduct independent DNA testing and that the jury should not accept the DNA results solely on the word of the State's witnesses.

The court further found that defense counsel did not move to suppress the DNA test results because he did not believe he had a valid legal basis for doing so. The court also found that Greenlee was surprised by Childs-Henry's testimony that there was DNA sample remaining, that Greenlee explored this issue with the prosecutor and the State's witnesses, and that Greenlee determined that there was in fact no DNA sample left to test. And the court found that the problems with the HPD Crime Lab did not become public knowledge until November 2002. The court further found that, against counsel's advice, applicant made an intelligent, knowing, and voluntary decision to testify. The court found that Greenlee believed that the negative consequences of applicant's decision to testify outweighed the negative impact of the DNA evidence.

Based upon its own independent recollection of the trial, the court found that Greenlee's cross-examination of HPD Crime Lab personnel was "effective trial strategy in that it was aggressive, probing, confrontational and effective" regarding an attack on the HPD Crime Lab protocol. The court also found that Greenlee's decisions — not to retain a consulting expert, not to pursue independent DNA testing, and not to seek a continuance" "were all made after a thorough investigation and were reasonable trial strategy." The court echoed this particular finding in its conclusions of law, and the court also concluded that applicant failed to show that the trial court would have committed error in denying a motion to suppress the HPD Crime Lab's DNA test results.

2. Deficient Performance

Strickland prescribes a two-part test for determining whether a defendant has been constitutionally deprived of effective assistance of counsel: (1) deficient performance, and (2) prejudice.

466 U.S. at 687, 104 S.Ct. 2052.

Deficient performance means that "counsel made errors so serious that counsel was not functioning as the `counsel' guaranteed the defendant by the Sixth Amendment." To establish deficient performance, "the defendant must show that counsel's representation fell below an objective standard of reasonableness." When trial counsel does not conduct a complete investigation, his conduct is "reasonable only to the extent that reasonable professional judgments support the limitations on investigation."

Id.

Id. at 688, 104 S.Ct. 2052.

Wiggins v. Smith, 539 U.S. 510, 533, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003) (internal quotation marks omitted).

Despite the habeas court's findings, the record raises serous issues with respect to Greenlee's performance. One reason Greenlee gave for declining to hire an expert was that the expert might validate the procedures followed by the HPD Crime Lab in this case. But defense counsel could have initially hired an expert for consulting-only purposes, to be designated later as a testifying expert if the expert's opinions turned out to be favorable. In Williams v. State, a case decided years before applicant's trial, we recognized that an indigent defendant was entitled to make his request for funds to hire an expert witness in an ex parte hearing. The rationale underlying our holding was the need to protect the confidentiality of the defendant's work product. In Skinner v. State, decided the same year as Williams, we explained that an expert's "comments about the strengths and weaknesses of the defense theory" were covered by the work product doctrine. Williams also quoted a Fifth Circuit discussion that emphasized the need to treat monied and indigent defendants equally:

958 S.W.2d 186, 194 (Tex.Crim.App. 1997).

Id.

956 S.W.2d 532, 538-39 (Tex.Crim.App. 1997).

The government should not be able to obtain a list of adverse witnesses in the case of a defendant unable to pay their fees when it is not able to do so in the cases of defendants able to pay witness fees. When an indigent defendant's case is subjected to pretrial scrutiny by the prosecutor, while the monied defendant is able to proceed without such scrutiny, serious equal protection questions are raised. . . .

A defendant with the resources to hire his own expert could hire someone on a consulting-only basis without the State ever discovering that he had done so. Under Williams and Skinner, an indigent defendant has the same capability.

See Pope v. State, 207 S.W.3d 352, 365-66 (Tex.Crim.App. 2006) (citing Williams and observing that a "simple solution" to the possibility that a prosecutor could comment on the existence of a defense expert is that a party can "investigate first, consult second, designate third").

In choosing to forgo an expert, defense counsel elevated the importance of his own personal investigation of DNA science. We will not discount the possibility that an attorney with no prior knowledge of DNA matters may become sufficiently knowledgeable through his own diligent sifting of the scientific literature, without the aid of an expert. But it is apparent in this case that Greenlee was not able to acquire sufficient knowledge on his own. Had he done so, he would have known that the HPD Crime Lab failed to adhere to accepted practices when it consumed all of the DNA evidence, and he would have understood that Chu's statistical frequency estimate was probably incorrect. Even without an expert of his own, Greenlee could have at least cross-examined Chu with relevant scientific literature — such as the FBI and NRC standards — regarding the desirability of retaining a portion of the DNA evidence for additional testing. Defense counsel could have pointed out, through questioning and argument, that two of the tubes of DNA material were redundant and did not have to be tested. That Greenlee did none of these things shows that he was probably ill-equipped to research the field without expert assistance.

See TEX.R. EVID 803(18) ("Learned Treatises" exception to the hearsay rule).

Even so, defense counsel should have conducted an investigation that was diligent enough to reveal that he needed the help of an expert in preparing for trial and cross-examining witnesses. In 2001, a variety of investigative tools existed that would have enabled Greenlee to determine that he needed an expert witness, including: an informal pre-hiring consultation with an expert, consultation with knowledgeable defense attorneys, continuing legal education, use of a medical library, and use of the internet. Although Greenlee consulted other attorneys, we cannot conclude that doing so was sufficient investigation in this case given the fact that Greenlee still lacked much understanding of DNA science.

And expert testimony would probably have given a boost to the defense beyond what could have been accomplished through cross-examination. A live expert could have explained that the FBI and NRC guidelines were not followed when Chu consumed virtually the entire sample in this case, and an expert could have testified to the flaws in Chu's statistical frequency calculation and given a more accurate calculation of the random match probability. A defense expert's testimony on these subjects would have been likely to carry more weight with a jury than cross-examination alone. Cross-examination may render unnecessary the presentation of one's own expert, especially if the cross-examination aggressively exploits learned treatises, but that did not happen in this case.

See id.

Greenlee did bring up the unfairness of consuming the DNA evidence, but Chu's rejoinder was that the sample size was so small that it had to be used up. The defense's cross-examination of Chu gave the jury no reason to think that Chu's effective consumption of the entire sample was anything other than an unfortunate but unavoidable circumstance.

The only evidence that potentially attacked Chu's explanation was the testimony elicited from Childs-Henry that the entire sample could be consumed if the chemist was sloppy. But Childs-Henry also said that the entire sample could be consumed if it was too small, which was consistent with Chu's explanation. And while Childs-Henry indicated that the sample was large enough to have some left for retesting, the credibility of that assessment was undermined by her error in believing that some sample still remained. Defense counsel did a creditable job of impeaching Childs-Henry with her inability to explain her own procedures; her mistake regarding the existence of any remaining sample, though a surprise, impeached her credibility as well.

However, we reject applicant's contention that Greenlee was deficient in failing to move for a continuance after Childs-Henry's testimony that DNA material remained to be tested. The evidence at trial (as well as the habeas proceedings) was that Childs-Henry was simply mistaken in her belief that there was liquid extract remaining. We do not think her testimony at trial fairly raised the notion that there might be trace amounts of DNA adhering to seemingly empty tubes. To the extent that her testimony suggested that there should have been DNA material remaining, that conclusion was one that Greenlee should have drawn prior to trial, and we have already concluded that Greenlee was deficient in that regard.

3. Prejudice

Having found deficient performance, we turn to the question of prejudice. The prejudice prong of Strickland requires showing "a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different." "A reasonable probability is a probability sufficient to undermine confidence in the outcome." It is not enough that counsel's errors could have had "some conceivable effect on the outcome of the proceeding," but a defendant does not have to show that counsel's deficient conduct "more likely than not altered the outcome of the case." This usual standard for showing prejudice is not always sufficient, however. If the proceeding was "rendered neither unreliable nor fundamentally unfair" by counsel's deficient performance, then the prejudice question can be answered in the negative to prevent the defendant from obtaining a "windfall."

466 U.S. at 694, 104 S.Ct. 2052.

Id.

Id. at 693, 104 S.Ct. 2052.

Lockhart v. Fretwell, 506 U.S. 364, 366, 113 S.Ct. 838, 122 L.Ed.2d 180 (1993).

As we explained in connection with his Youngblood claim, applicant would not have been entitled to the suppression of the DNA test results. Had counsel conducted a proper investigation, an attack might have been made on the quality of Childs-Henry's testing, but there is no indication that such an attack would do anything to undermine the validity of Chu's results, which were based upon his own testing. Defense counsel would have been able to introduce evidence that consuming the entire sample was bad laboratory practice, but there is no indication that the test results themselves flowed from any flawed practices. To the contrary, Bromwich characterized Chu's work as generating "good quality and clear results" and developing "clean, interpretable profiles." And a jury would still be confronted with the fact that the genetic typing occurred before applicant even became a suspect — eliminating the possibility of contamination from applicant's own DNA.

Had defense counsel conducted a proper investigation, he would have been able to introduce evidence that Chu's statistical frequency estimate was flawed, and he would have been able to introduce a correct estimate in line with Bromwich's calculations: e.g. a random match probability for the African-American population of 1 in 232,000. A probability of 1 in 232,000 is still quite low, although, with a city the size of Houston, it might allow for the possibility of a handful of other people matching the test results.

But when a 1 in 232,000 random match probability is combined with the other evidence introduced at applicant's trial, there is no reasonable probability of a change in outcome. The victim picked both applicant and his car out of separate lineups. Although the victim could not or did not identify applicant via closed-circuit television at trial, the victim indicated that he was positive when he picked applicant out of a pretrial lineup, and the circumstances surrounding that pretrial identification (the victim's smile when the camera panned on applicant) further reinforces the identification. A composite sketch drawn from the victim's descriptions was consistent with applicant's appearance. The exterior and interior of applicant's house, along with his yard, were consistent with the descriptions given by the victim to law enforcement. Purple clothing found in applicant's house and car were also consistent with the victim's prior descriptions, as was a dark baseball cap recovered from applicant's car. The color of the interior of applicant's car was consistent with the victim's description, and the exterior of the car was damaged in a way that was consistent with the victim's brother's description. And the bottle of orange cocoa butter lotion found in applicant's house matched the victim's description of "orange grease" used by the perpetrator.

At trial, E.T. identified from photographs various items that looked like items he encountered when he was kidnapped: the outside of applicant's house, the brown couch, the purple clothes, and a Houston Oilers cannister. E.T. pointed to circles on applicant's sidewalk that he recognized.

Applicant's defensive evidence was not helpful. His alibi was full of holes. Although electronic monitoring established that applicant arrived home at 3:07 p.m., that arrival time did not exclude him as the perpetrator. The family members who attempted to establish an alibi gave testimony that was inconsistent with each other and with their own prior statements, resulting in estimates for applicant leaving the Savage residence ranging from 1:30 to 3:30 p.m. The prosecution effectively exploited David Savage's testimony that he saw applicant on television on February 12th, at which time applicant was not a suspect, but the composite drawing was being shown. Finally, applicant himself stated that the victim was taken to an apartment "before he come to my house." It is not apparent how this could be taken as anything other than an inadvertent admission that the child was at applicant's house.

The jury also heard applicant's refusal to acknowledge responsibility for any of the offenses he had already been convicted of. And applicant's testimony, some of which was evasive, showed his disregard for the geographical-limitation conditions of his parole and his numerous violations of those conditions.

Applicant's testimony also suggested his own consciousness of guilt, as he advanced implausible explanations in an attempt to distance himself from evidence that tended to link him to the crime: He denied owning the cocoa butter lotion — saying that he must have taken it from his brother's room and that it was owned by his brother's sister. And applicant denied that his car was damaged on February 11th, claiming that the police must have caused the damage while the car was in their custody. And he accused parole officer Butler of lying about the conditions of his parole. All of the evidence shows a very strong case against applicant, even when Bromwich's criticisms of the HPD Crime Lab's conduct are taken into account.

Although not argued by applicant, a position could perhaps be taken that applicant might not have testified or might not have insisted on the parole officers' testimony absent Greenlee's deficient performance on the DNA matters. Such an argument would not change our conclusion with respect to prejudice for at least three reasons. First, even accepting Bromwich's criticisms, and even without the testimony of applicant or his parole officers, the evidence so strongly supports conviction that we would not find a reasonable probability that the outcome would be different. Second, effectively presenting Bromwich's criticisms to the jury would not change the apparent need for an alibi, so we cannot find a reasonable probability that applicant would have refrained from testifying or would have refrained from insisting on the testimony of his parole officers. Third, even if it could be shown that applicant would not have testified, it would be unclear whether he could avoid the incriminating effects of his testimony under the prejudice prong of Strickland. The Seventh Circuit and the Supreme Court of Indiana have suggested that, under Lockhart v. Fretwell, a defendant's own incriminating testimony, at least where it was gratuitous, can defeat a finding of prejudice. United States v. Parker, 609 F.3d 891, 895-96 (7th Cir. 2010) ("Parker has only himself to blame for admitting under oath to a quantity of drugs he now disputes."); Smith v. State, 689 N.E.2d 1238, 1248 (Ind. 1997) (At sentencing, defendant testified that "he was no longer claiming that he did not molest his daughter." Having "admitted that his trial testimony to the contrary was not truthful," the defendant could not "on appeal make a credible argument that the result of his trial was unjust and unreliable."). No one forced applicant to testify that the complainant was taken to his house, that the cocoa butter lotion found at his house was not his, that one of his parole officers lied, that applicant had no regularly scheduled reporting date, or that he did not commit the sex offenses for which he had been previously convicted.

Finally, we conclude that the new test results from ReliaGene and Orchid Cell-mark do not alter the prejudice analysis in applicant's favor. It is not a foregone conclusion that an "effective" attorney would have sought further testing upon discovering the errors committed by the HPD Crime Lab in this case. Even so, we have already explained, in connection with the Youngblood discussion, that the test results from ReliaGene and Orchid Cellmark were not exculpatory, but rather tended to incriminate applicant.

See Skinner v. State, 293 S.W.3d 196, 202-03 (Tex.Crim.App. 2009) ("counsel explained that he did not ask for testing because he was afraid the DNA would turn out to be appellant's").

III. CONCLUSION

For the reasons outlined above, we reject applicant's Youngblood, perjury/false statement, and ineffective assistance of counsel claims. We deny relief.

Applicant also advanced two claims in his habeas application that we did not file and set: a claim of actual innocence and a Brady claim. Various aspects of our discussion defeat these claims as well. We reject these claims without further comment.

COCHRAN, J., joined except section II B 2.

PRICE, JOHNSON, and HERVEY, JJ., concurred.