dissenting:
Reluctantly and respectfully I must dissent for the third time in this case.
Eleven courts have considered various aspects of petitioner Mata’s claims. Only this court, on three separate occasions, has responded favorably.
Mata was convicted of first-degree murder in the Superior Court of Kern County, California. He appealed his conviction to the California Court of Appeal for the Fifth District, where he alleged for the first time that his constitutional rights were violated by the pretrial identification procedures discussed in the majority opinion. The California Court of Appeal affirmed the conviction, and made specific findings of fact concerning Mata’s claims. Mata filed a petition for a writ of habeas corpus in the Superior Court of Marin County, California. That court denied the petition, as did the California Court of Appeal for the Fifth District and the California Supreme Court.
Finally, Mata sought collateral review in the federal courts. The district court refused to issue a writ of habeas corpus. This court twice gave judgment in favor of Mata, 611 F.2d 754 (9th Cir.1979); 649 F.2d 713 (9th Cir.1981), and twice had its opinion vacated and remanded by the Supreme Court, with instructions to comply with the requirements of 28 U.S.C. § 2254(d). Sumner v. Mata, 449 U.S. 539, 101 S.Ct. 764, 66 L.Ed.2d 722 (1981) [“Mata I”]; Sumner v. Mata, 455 U.S. 591, 102 S.Ct. 1303, 71 L.Ed.2d 480 (1982) [“Mata II”].
My present dissent, like its immediate predecessor, is shaped in part by the Supreme Court’s immediately preceding remand. Thus, I now dissent because in my view the findings of fact in the state proceedings before the California Court of Appeal are “fairly supported by the record” as required by 28 U.S.C. § 2254(d)(8). Also I continue to insist that the pretrial identification procedures were not. impermissibly suggestive and that any error in the identification of Mata that might have existed was harmless. Finally, I reject the majority’s view that the failure to hold a lineup deprives the petitioner of his constitutional due process rights.
I.
THE APPLICATION OF SECTION 2254(d)(8)
A. Section 2254(d)
To minimize the friction that inevitably has resulted from federal habeas corpus review, and to show respect for principles of comity and federalism, Congress, as the Supreme Court explained in Mata I, in 1966 enacted 28 U.S.C. § 2254(d) as an amendment to the Federal Habeas Act of 1867. 449 U.S. at 550, 101 S.Ct. at 770, 66 L.Ed.2d 722; Suggs v. LaVallee, 570 F.2d 1092, 1112 (2d Cir.), cert. denied, 439 U.S. 915, 99 S.Ct. 290, 58 L.Ed.2d 263 (1978). Under section 2254(d), as the Supreme Court stated in Mata I, and repeated emphatically in Mata II, a federal court engaged in habeas corpus review of a state decision must presume the findings of fact made in the state court system to be correct.
Two exceptions to this presumption exist. First, the federal court may disregard the state findings of fact if the habeas corpus petitioner can prove “by convincing evidence that the factual determination by the State court was erroneous.” 28 U.S.C. § 2254(d). Proof by “convincing evidence” is a burden more rigorous than the “preponderance of the evidence” standard. 449 U.S. at 551, 101 S.Ct. at 770. Second, the federal court may defeat the presumption of correctness if it finds that the requirements of section 2254(d)(l)-(7) or section 2254(d)(8) are satisfied. If the federal court bases its decision on this second exception, it is obliged to explain specifically the reasons for its conclusion. 102 S.Ct. at 1307; 449 U.S. at 552, 101 S.Ct. at 771.
Moreover, a finding by the reviewing court that one of these exceptions applies to a particular finding of fact does not relieve that court of the duty to defer to the state *1257court’s other findings of fact. The federal court is not free to put aside its obligation to defer to these other findings merely because another particular state finding is held not to be fairly supported by the record. The majority recognizes this in Part III of its opinion. See also Harris v. Pulley, 692 F.2d 1189, 1199 (9th Cir.1982).
B. Exceptions to Section 2254(d)
It is clear from Mata II that if the facts found in the Fifth District Opinion are correct, there is no support for the majority’s conclusion that a deprivation of due process rights has taken place. See 102 S.Ct. at 1306-07 & n. 6. Thus, the majority in the final analysis must rely on an exception to section 2254(d) to support its holding.
The “convincing evidence” exception to section 2254(d) cannot apply here. Five courts have found that the state findings of fact was correct, and the Supreme Court has twice had the opportunity to uncover such “convincing evidence,” but has declined to do so. Moreover, the majority has not pointed to any “convincing evidence” in the three times that this case has been before us; indeed, on the previous remand, the majority found that it was “substantially” in agreement with the state court’s factual findings. 649 F.2d at 717.
The majority therefore looks to section 2254(d)(8) to escape the presumption that the Fifth District’s findings of fact are correct. Majority Opinion at 1251. Under this section the presumption of correctness of state court findings of fact can be disregarded where the federal court, after a review of the record of the state proceeding, concludes that the record “does not fairly support” the state findings. No other basis for refusing to defer to state findings of fact is available to the majority.
The majority appears to assume that a record “does not fairly support” the state court when a federal reviewing court disagrees with the state court’s interpretation of the record. See Majority Opinion at 1250-1251. Such a construction fails to give due deference to a state court’s factual determination and to reflect the intent of Congress in enacting section 2254(d)(8). See pp. 1256-1257 supra.
Section 2254(d) was promulgated in response to Townsend v. Sain, 372 U.S. 293, 83 S.Ct. 745, 9 L.Ed.2d 770 (1963), and it is to that case that one must turn when seeking the meaning of the statute. See Cuyler v. Sullivan, 446 U.S. 335, 341-42, 100 S.Ct. 1708, 1714-15, 64 L.Ed.2d 333 (1980); see also 649 F.2d at 716 n. 6. Townsend refers us to Blackburn v. Alabama, 361 U.S. 199, 80 S.Ct. 274, 4 L.Ed.2d 242 (1960), and Fiske v. Kansas, 274 U.S. 380, 47 S.Ct. 655, 71 L.Ed. 1108 (1927), for the meaning of “fairly support.” See 372 U.S. at 313, 316, 83 S.Ct. at 757, 758. According to these cases, a finding is not fairly supported by the record if it is “shown by the record to be without evidence to support it.” 274 U.S. at 385, 47 S.Ct. at 656. See also 361 U.S. at 208-09, 80 S.Ct. at 280-81 (a finding “without any evidence in the record” is not fairly supported by it).
Thus, to apply section 2254(d)(8) properly one must ask whether there is any evidence in the record to support the state court’s finding of fact. Moreover, when the record is searched for support for the state findings, it embraces the entire record. It is not limited by the text of the state opinion, as the majority assumes. Majority Opinion at 1250-1251. See Townsend v. Sain, 372 U.S. at 314-15, 83 S.Ct. at 757-58; Fowler v. Jago, 683 F.2d 983, 987 (6th Cir.1982). If any evidence that fairly supports the state finding can be found in the record, the federal court must presume the state finding to be correct, in the absence of invocation of the “convincing evidence” exception previously discussed.1 See United States ex rel. Jones v. Franzen, 676 F.2d 261, 264—65 (7th Cir.1982). See also United States v. Warden, Illinois State Penitentiary, 566 F.2d 28, 30 (7th Cir.1977) (a federal court *1258should defer to a state court finding of fact, “particularly when a state appellate court has made an independent factual review of an adequate record”); Hardeman v. California, 445 F.2d 258, 259 (9th Cir.) (Ely, J.) (“A full and fair evidentiary hearing on this was conducted by the California courts, and the District Court was thus entitled to accept the factual determination of the state courts”); cert. denied, 404 U.S. 998, 92 S.Ct. 572, 30 L.Ed.2d 550 (1971).2 Finally, to invoke section 2254(d)(8) to overcome the presumption of correctness, a federal court must give a satisfactory explanation for its decision. Mata II, 102 S.Ct. at 1307.
In the remainder of this portion of my dissent I shall analyze the record in a manner consistent with these principles.
C. The Record Fairly Supports the State Court’s Findings
The Supreme Court in Mata II distinguished between the underlying “historical facts” and the application of law to those facts in considering a habeas corpus petition under 28 U.S.C. § 2254(d). 102 S.Ct. at 1306-07. Making this distinction is not always easy and infrequently can be done without some dispute. However, it is legitimate to characterize the following key sentence of the California Court of Appeal’s opinion as a recitation of findings of “historical facts”:-
We find further that there is no showing of influence by the investigating offieers[;] that the witnesses had an adequate opportunity to view the crime; and that their descriptions are accurate.
Excerpt of Record at 82-83. The record clearly supports each of these three findings.3
1. No Showing of Influence by the Investigating Officers
The record establishes that Almengor was not subjected to any pressure to make an *1259identification of the assailants. Lieutenant King testified that Almengor was not told that he was a suspect in the slaying, and that Almengor was never threatened in any way with being charged with the crime. R.T. at 764-65. The only promise made to Almengor was that he would be protected, and the possibility of withholding Almengor’s parole was not a consideration when he was interrogated. R.T. at 726 — 28. Almengor confirmed that he was promised nothing and was not threatened by the investigating officers. R.T. at 751.
Allen’s testimony also shows that he was neither promised anything nor threatened by the investigating officers. R.T. at 649-50, 800. Parole considerations did not affect his actions. R.T. at 969. Allen’s initial hesitation about participating in the investigation of the murder was caused by his reluctance to get involved. Allen testified:
Q: Now, when you were telling the correctional officers the Sheriff’s officers that you couldn’t identify anybody, and they showed you pictures that you couldn’t identify, and you couldn’t identify anybody, were you telling the truth to them, or were you just trying to not get involved?
A: Just trying to not get involved.
Q: When you finally identified three people from pictures, like you told us before, were you at that time doing your honest best to identify the people who actually were involved?
A: Yes, sir.
Q: Did the correctional officers or the Sheriff’s officers try to tell you or attempt to tell you in any way who you should identify?
A: No. They just had a bunch of pictures laid out.
R.T. at 650.
The majority admits that the array from which the identifications were made was composed of photographs that were “similar in terms of size, color, and pose. All of the inmates pictured were of Mexican descent.” Majority Opinion at 1249. There was no allegation that any suggestive procedures were employed during the identification sessions.
The majority argues, however, that Almengor and Allen were influenced by seeing the prison authorities remove the three defendants from segregation to have their photographs taken. Almengor’s testimony shows that he did not see Mata on his way to the photographer, R.T. at 444-46, and Allen could not recall saying to the defendants’ private investigator that he had witnessed the incident. R.T. at 634. Moreover, the California Court of Appeal found that the record does not show that Almengor and Allen were influenced in any way by the incident, even assuming that they had seen it.
2. The Witnesses Had an Adequate Opportunity to View the Crime
Almengor had an excellent opportunity to see both Mata and the crime itself. Almengor testified that he was “reasonably close” to Mata at the time of the murder, R.T. at 391, was opposite him, R.T. at 742-43, and looked him in the face. R.T. at 382. Almengor said that he saw Mata stab at the victim, R.T. at 393, and “sloop over.” Id. *1260Almengor then fought hand-to-hand with Mata “all the way up the dorm.” R.T. at 381, 393, 453-56.
Allen had less of an opportunity to see Mata, since Vargas had attacked Allen when he tried to intervene in the assault. R.T. at 624-26. But Allen did get some view of the three attackers, R.T. at 586, 593-94, and was able to describe their appearance. R.T. at 621-22.
3. The Witnesses’ Descriptions Were Accurate
Almengor described the height, weight, build, complexion, and clothes of Mata. R.T. at 375-76, 419, 969-71. See note 11 infra. Almengor also gave a description of the details of the crime, including Mata’s exact movements during the entire assault. R.T. at 391-93. Allen also described the clothes of the assailants, as well as the attack, and had seen enough of the fight to recognize the attackers later by their photographs. R.T. at 585-94, 801, 1117. Other testimony confirmed details of these descriptions. E.g, R.T. at 510, 658-62, 1176-77. The California Court of Appeal found that the witnesses’ descriptions were accurate, the only evidence to the contrary being the testimony of defendants’ private investigator concerning a possible variation in two details of Allen’s testimony. R.T. at 1117-18.
4. Conclusion
I conclude that each of the three findings of the California Court of Appeal is supported by the record. The presumption of correctness provided by section 2254(d) attaches to these findings. The majority errs in not recognizing this. See Townsend v. Sain, 372 U.S. 293, 313-16, 83 S.Ct. 745, 757-59, 9 L.Ed.2d 770 (1963).
II.
SUGGESTIVE IDENTIFICATION CLAIM
A. An Application of the Underlying Facts Does Not Establish a Due Process Violation
As the Court put it in Manson v. Brathwaite, 432 U.S. 98, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977), “reliability is the linchpin in determining the admissibility of identification testimony .... ” Id. at 114, 97 S.Ct. at 2253. But while “reliability” may be the inquiry’s “linchpin,” a reviewing court must remember that, as a “linchpin” serves to keep the wheel from slipping off the axletree, so “reliability” is but the key to ensuring that no due process violations occur. The fundamental issue remains whether the facts establish a due process violation. See Part IV infra.
The reliability of an identification is determined by reference to the five factors set out in Neil v. Biggers, 409 U.S. 188, 199-20, 93 S.Ct. 375, 382-83, 34 L.Ed.2d 401 (1972). These include “the opportunity of the witness to view the criminal at the time of the crime, the witness’ degree of attention, the accuracy of the witness’ prior description of the criminal, the level of certainty demonstrated by the witness at the confrontation, and the length of time between the crime and the confrontation.” See Mata II, 102 S.Ct. at 1307 n. 10; Manson, 432 U.S. at 114, 97 S.Ct. at 2253. I shall consider each of these factors.
1. Adequate Opportunity to View
The California Court of Appeal found that the witnesses had an adequate opportunity to view the “crime,” and, as pointed out above, there is evidence in the record to support that finding. The majority contends that the Neil v. Biggers test refers to the opportunity to view the “criminal,” not merely the “crime.” Assuming, arguendo, the majority is correct, it does not follow that the state court finding is entitled to no deference. It is entitled to deference if it is supported by the record as a whole. See Townsend v. Sain, 372 U.S. at 314-15, 83 S.Ct. at 757-58. The record shows that both Almengor and Allen saw the three men involved in the attack as well as the crime itself. R.T. at 392-93, 586.
Admittedly, Allen could not see all three assailants well. R.T. at 586. But the majority’s reference to Allen’s statement after *1261the assault that he saw nothing should be viewed in the context of Allen’s initial refusal to talk to the authorities about the crime. See R.T. at 631, 649-50.
The majority concedes, however, that Almengor’s opportunity to view Mata was “adequate,” but argues that adequacy is “a finding hardly specific enough upon which to rely in reaching a conclusion that must be based on the totality of the circumstances.” Majority Opinion at 1252. This will not do. Neil v. Biggers requires no more than an “adequate opportunity.” If the opportunity was adequate, that factor in the test is satisfied. Such satisfaction cannot also indicate a violation of Mata’s due process rights. See Mata II, 102 S.Ct. at 1306 n. 6.
2. Degree of Attention
The majority is correct when it states that Almengor paid a great deal of attention to the assailant he identified as Mata, but that Allen focused his attention on the attacker in front of him.
3. Accuracy of the Prior Description
The California Court of Appeal found that the descriptions of both witnesses were accurate. There is evidence in the record to support that finding. R.T. at 375-76, 391-93, 419, 510, 586-94, 621-24, 741-42, 801-02.
4. Level of Certainty at Confrontation
The state court found that Almengor made three “possible” identifications on October 19 — not “positive” identification, as the majority claims. Majority Opinion at 1252-1253. And since no photograph shown to the witness before October 30 resembled Mata’s appearance at the time of the assault, the witnesses understandably were “uncertain,” id., as the state court indicates.
The state court found that both Almengor and Allen identified Mata on October 30 — the first time that the witnesses were shown a recent photograph of him. See my first dissent, 611 F.2d at 761 (Sneed, J., dissenting). The record shows that the October 30 identifications of Mata were “positive.” R.T. at 718-22.
5. Length of Time Between the Crime and the Confrontation
The California Court of Appeal found that eleven days passed between the date of the crime and that of the identification of Mata. The majority concedes that the lapse of a week does not render an identification unreliable. Nor does eleven days. See United States v. Hammond, 666 F.2d 435, 440 (9th Cir.1982). See also United States v. Scriber, 499 F.2d 1041, 1046-47 (D.C.Cir.1974) (identification 11 days after crime is reliable); United States v. Kulp, 365 F.Supp. 747, 758-59 (E.D.Pa.1973), aff’d mem., 497 F.2d 921 (3d Cir.1974) (“several weeks”).
6. Summary
This examination of the Neil v. Biggers test in the light of the section 2254(d) presumption of correctness shows that the Neil v. Biggers test is satisfied, and that the identifications are reliable. The California Court of Appeal found that both Almengor and Allen had an adequate opportunity to view the crime and described the assailants accurately. Both witnesses made positive identifications of Mata at the first opportunity to do so, and the identification took place reasonably soon after the crime. Almengor, it is true, paid much more attention to Mata than did Allen, but that should not render Allen’s testimony unreliable. United States v. Hammond, 666 F.2d at 440 & n. 2.
B. There is No Need to Balance the “Corrupting Effect” of the Identification Procedure in this Case
The majority weighs the reliability of the identifications against “the corrupting effect of the photographic identification procedures employed by the prison authorities.” Majority Opinion at 1253. The majority bases this balancing test on the language of Manson v. Brathwaite, 432 U.S. 98, 114, 97 S.Ct. 2243, 2253, 53 L.Ed.2d 140 (1977). The majority, however, misunderstands the “corrupting effect” test.
The Manson Court held that the reviewing court must weigh against the Neil v. Biggers factors “the corrupting effect of *1262the suggestive identification itself.” Id. (emphasis added). Weighing is not required when the identification is not “suggestive” although prudence suggests that weighing might well accompany all identifications. “Suggestive identifications” can occur in the case of a single photograph shown to a witness for identification, id. at 116, 97 S.Ct. at 2254, a photographic array designed to make the identification the result of impressions gained during the identification process rather than during the crime, United States v. Hanigan, 681 F.2d 1127, 1133 (9th Cir.1982), or a one-person “showup.” Styers v. Smith, 659 F.2d 293, 297 (2d Cir.1981). In contrast, the Manson Court expressly approved a photographic array including a reasonable number of persons who resemble the suspect, 432 U.S. at 117, 97 S.Ct. at 2254, precisely the type of array used in the identification of Robert Mata. See 611 F.2d at 756.
Nonetheless, the majority finds that the identification procedures in the present case were “suggestive.” The majority makes six allegations to support its finding of suggestiveness. These are (1) that Almengor was shown photographs on three occasions, (2) that photographs previously identified by Almengor as “possible” were removed from the array, (3) that the array was reduced in size, (4) that the defendants’ photographs were the only ones updated for the October 30 array, (5) that Almengor and Allen saw Mata being taken out to be photographed, and (6) that the prison authorities put pressure on Allen to cooperate. Majority Opinion at 1255. The finding of the majority is impermissible under a proper reading of § 2254(d).
To begin with, a finding of suggestiveness or no by a state court is a mixed question of law and fact, to which the reviewing court must defer unless an exception to section 2254(d) is applicable. Mata II, 102 S.Ct. at 1306-07. The state court here found the identification to be fair and not suggestive. No exception to section 2254(d) is available. The record supports the state court’s findings.
First, Almengor was shown a photographic array on three occasions because, as it turned out, there were no up-to-date, and therefore identifiable, photographs in the earlier arrays of the men later picked out by Almengor as the assailants. R.T. at 595-96, 721-22. United States v. Higginbotham, 539 F.2d 17, 23 (9th Cir.1976), approves the showing of two different photographic arrays to a witness, while condemning “repeated” showings of the same photographic display in the hope that the witness eventually will identify someone — a situation not alleged here. See also United States v. Brown, 461 F.2d 134, 142-43 (D.C. Cir.1972) (en banc) (three arrays and a lineup before identification made is not suggestive).
Second, the two photographs identified by Almengor as “possible” were not included in the October 30 array because the persons identified could not have participated in the assault. Keeping the photographs in the array would not have advanced the search for truth.
Third, the reduction in the size of the photographic array is irrelevant, as long as the size is “reasonable,” as this array was. Manson, 432 U.S. at 117, 97 S.Ct. at 2254.
Fourth, although the three defendants in this case were the only ones to have their photographs updated for the October 30 array, all of the photographs displayed on that day were reprinted so as to appear identical. R.T. at 716.
Fifth, Almengor did not see Mata being taken out to be photographed, R.T. at 444— 45, and Allen may not have seen the incident. R.T. at 634. Moreover, the California Court of Appeal found that there was no showing by Mata that Allen was influenced by having seen Mata on his way to the photographer.
Finally, there is the question of undue police influence on Allen. The state court found that there was no showing of influence by the investigative officers on Almengor and Allen, and the record supports that finding. See Part I, supra; 611 F.2d at 760-61 (Sneed, J., dissenting). The federal court should presume this underlying finding of fact to be correct. Mata II, 102 S.Ct. at 1306-07.
Thus, I would acknowledge that the identification procedures employed by the inves*1263tigative officers here were not, for the purposes of habeas corpus proceedings under section 2254, unduly “suggestive.” Therefore these procedures had no “corrupting effect” that requires the weighing mandated by the Manson Court. However, a weighing of the non-suggestive procedures, as prudence dictates, against the Neil v. Biggers factors only strengthens my conclusion that the majority erred in granting Mata’s petition. No due process violation' occurred.
III.
ANY ERROR WAS HARMLESS
Finally, any error was harmless, beyond a reasonable doubt. The federal district court so held.4 The majority disregards that holding,5 and finds that error was not harmless. Majority Opinion at 1255-1256. By so doing, the majority ignores the strongly inculpating evidence against Mata apart from the testimony of Allen, overestimates the dimensions of any error in Almengor’s identification of Mata, and fails to evaluate properly the role of the jury in evaluating the evidence against Mata.
First, the majority does not mention that the testimony of the “third witness,” Paul Childress, an inmate who knew Mata, positively identified Mata as taking part in the murder.6 There was also testimony about a Mexican Mafia plot involving Mata to kill *1264Arias, the victim.7 And other testimony-indicated that Mata had shown the testifying witness how Mata had “hit” the victim.8 Moreover, a blood-stained jacket and knife were found in a locker near Mata’s bed. R.T. at 658-62.
Second, the only meaningful defect in Almengor’s identification of Mata is that Almengor did not pick Mata out of a photographic array until eleven days after the murder.9 But Almengor’s identification occurred the first time that he saw a recent photograph of Mata. Unlike Allen, Almengor did not see Mata being led out to have his photograph taken,10 and there is no allegation that Almengor was subjected to any pressure on the part of the prison authorities to make an identification. As the district court put it, the identification of Mata by Almengor was “strong.” Excerpt of Record at 96.
Finally, the majority cites to United States v. Field, 625 F.2d 862 (9th Cir.1980), for the proposition that any error in the identification of Mata could not be harmless because “we do not know which witness the jury may have believed.” Majority Opinion at 1255. The majority misreads Field and the evidence in the present case.
In Field, the defendant was convicted on the basis of identifications by three witnesses. 625 F.2d at 865. The court held that two witnesses were subjected to unconstitutionally suggestive identification procedures and that the testimony of the third was highly unreliable. Id. at 668-70. The court concluded that any error could not have been harmless since there was no reliable witness who had identified the defendant as the criminal whom the jury could have believed.
*1265That is not the case here. The witness Childress made a positive identification of Mata, and Almengor’s testimony, especially taken in the context of the circumstantial evidence against Mata, is very strong. Allen’s identification may be suspect, but his testimony is merely cumulative. Moreover, the majority says at most that Allen identified Mata out of fear of retaliation by the prison authorities, not out of a desire to cover up for someone else. See R.T. at 801. Such fear, even if it existed, provides no reason to assume that Allen’s identification of Mata was incorrect.
Given the strength of the case against Mata, I conclude that the district court was correct in holding that any error is harmless beyond a reasonable doubt.
IV.
FAILURE TO HOLD A LINEUP
The majority holds that federal courts in this Circuit should consider police use of photographic arrays instead of lineups as a factor which “bears on the extent of the corruptive effect of the identification procedure” when deciding whether a pretrial identification violates a suspect’s constitutional rights. Majority Opinion at 1254. According to the majority, this court should “consider the fact that the officers chose to use a less reliable identification procedure (e.g., photographic array) over a normally more reliable procedure (e.g., corporeal line-Majority Opinion at 1254 (emphasis in original). Presumably, such use suggests at least the possibility of a due process violation. up):
However, it is plain that an accused has no constitutional right to a lineup rather than a photographic array. See United States v. Robertson, 606 F.2d 853, 857 (9th Cir.1979). Nonetheless, the majority seeks to accomplish indirectly what the law explicitly rejects.
The majority effectively would force the police to abandon arrays in favor of lineups whenever possible. As Justice Stevens pointed out in Mata II, the issue the majority resolves has been “lurking in the background of this case.” 102 S.Ct. at 1308 (Stevens, J., dissenting). I continue to dissent from this needless intervention into the identification procedures employed by the police.
I explained in my first Mata dissent, 611 F.2d at 760-61, that requiring the police to employ a lineup when investigating a crime committed in a prison would place a terrible strain on prison resources.11 The majority now compounds the problem by making its principle apply with equal force outside the prison walls.
To the majority the price of due process compliance is unrelated to its substance. Majority Opinion at 1253-1255. In terms of logic the majority is correct. In terms of practical judgment the majority is wrong. *1266The courts cannot fashion rules to govern police conduct in a vacuum. We must take into account the dangers that both police and prison authorities face, and the role that those groups play in preserving the fabric of society in shaping the contours of constitutional due process. See, e.g., Terry v. Ohio, 392 U.S. 1, 22-24, 88 S.Ct. 1868, 1880-81, 20 L.Ed.2d 889 (1968). No judge with whom I am acquainted would assert, when pressed, to the contrary.
The Supreme Court has never said that identification by means of a photographic array including “so far as practicable ... a reasonable number of persons similar to any person then suspected,” Manson v. Brathwaite, 432 U.S. at 117, 97 S.Ct. at 2254 (citation omitted), is contrary to constitutional due process. Rather, the Manson Court recommended the use of an array substantially identical to that employed in the identification of Robert Mata. Id. The Court in Simmons v. United States, 390 U.S. 377, 386 n. 6, 88 S.Ct. 967, 972 n. 6, 19 L.Ed.2d 1247 (1968), did suggest in dictum that some types of photographic arrays might be less accurate than a lineup, but the Simmons Court did not find a due process violation even with an array of just six photographs. Id.12
The fact remains that the Supreme Court since Simmons has never forbidden the use of an array of photographs for identification, has never held unconstitutional such an identification procedure, and has never held that the use of an array constitutes a suggestive identification technique having a corrupting effect on identification testimony. See, e.g., Tibbs v. Florida, - U.S. -, 102 S.Ct. 2211, 2213 n. 2, 72 L.Ed.2d 652 (1982); United States v. Ash, 413 U.S. 300, 93 S.Ct. 2568, 37 L.Ed.2d 619 (1973).
The majority’s holding therefore goes far beyond the Supreme Court’s photographic array decisions. It beckons to the Supreme Court to follow after its lead; but I had always thought our duty was to follow the way marked out by the Supreme Court, not to lead it, even when our way might appear to us to be the true path of righteousness.
. Piche v. Rhay, 422 F.2d 1309 (9th Cir.1970), the only Ninth Circuit case to address this issue, noted that a federal court must accept the findings of a state trier of fact if the state findings are reliable — that is, “supported by substantial evidence.” Id. at 1311. While Piche might have adopted a more rigorous standard of review than that of Townsend v. Sain, it is irrelevant for the purposes of this *1258case, since there is more than “substantial evidence” to support the Fifth District Opinion.
. In Lonberger v. Jago, 651 F.2d 447 (6th Cir.1981), a case similar to the present one, the court concluded that the state court’s finding that the petitioner’s guilty plea was involuntary was not fairly supported by the record. The court did not, however, seek support for the state court finding before reaching its decision. The Supreme Court has granted certiorari. 454 U.S. 1141, 102 S.Ct. 998, 71 L.Ed.2d 292 (1982).
. The Fifth District Opinion is a 16-page document covering a number of constitutional and evidentiary challenges to the convictions of Mata and his co-defendants. The relevant sections, annotated with references to the record, state:
I. The Photographic Lineup Procedure.
Both appellants Gallegos and Mata contend that there were prejudicial errors in the pretrial identification of them by certain prosecution witnesses.
Three inmate witnesses testified that they saw the stabbing take place. All three— Childress, Almengor, and Allen — identified all three defendants; but their testimony is attacked by appellants Gallegos and Mata on the basis of claimed improper pretrial photographic identification procedures. The witnesses were shown a number of photographs of Tehachapi inmates in an attempt to identify the slayers. Almengor was interviewed and shown photos on October 19, 1972, the day of the incident. [R.T. at 397, 422], He made a possible identification of appellant Vargas [R.T. at 423], but made possible misidentifications of the other two participants. [R.T. at 712-13, 766], On October 30, 1972, more recent photos were presented to Almengor and he identified all the appellants. [R.T. at 402, 718]. On October 27, 1972, Allen was shown photographs but stated he could not make an identification because the photographs were old. [R.T. at 595, 721], On October 30, 1972, more photos were presented to Allen and he identified all three appellants. [R.T. at 722]....
Appellants argue that the witnesses Almengor and Allen were housed in the same segregation unit with appellants, that they were aware that appellants were removed from the segregation unit to have their pictures taken and that this makes their identification inadmissible. But they make no showing, and the record supports none, that the witnesses were in fact influenced in their identifications by this action of the investigating officers.
... [A] violation of due process occurs and a conviction will be set aside only if the photographic identification procedure was so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification. (See Simmons v. United States). . . .
Reviewing the facts of the present case to determine if the particular photographic identification procedure used contained the proscribed suggestive characteristics, we first find that the photographs were available for cross-examination purposes at the trial. *1259[R.T. at 712-17]. We further find that there is no showing of influence by the investigating officers [R.T. at 649-50, 726-28, 751, 764-65, 800, 953, 968-69] [;] that the witnesses had an adequate opportunity to view the crime [R.T. at 381-82, 391-93, 439, 453-56, 586, 593-94, 624-26, 741-42, 801-02]; and that their descriptions are accurate. [R.T. at 375-76, 391-93, 419, 510, 586-94, 621-24, 658-72, 801, 969-71], The circumstances thus indicate the inherent fairness of the procedure, and we find no error in the admission of the identification evidence. Excerpt of Record at 81-83 (emphasis in original). The Fifth District also found in confirmation of the eyewitness testimony that:
... Eye witness testimony placed all of the appellants at the scene of the murder and participating in it. [R.T. at 279, 375-83, 510, 587]. Physical evidence, including gloves, a knife, and a jacket with blood on it were found. [R.T. at 658-61, 670-72].
Id. at 92.
. The district court also found that any irregularities in the pretrial identification of Robert Mata did not amount to a violation of his constitutional rights.
The relevant portion of the district court opinion states: “On the merits, the Court concludes the petition must be denied. While petitioner has established that irregularities occurred in the pre-trial photographic identification of petitioner by witnesses Almengor and Allen, these irregularities did not so taint the in-court identifications of petitioner by these witnesses as to establish a constitutional violation under Simmons v. United States, 390 U.S. 377, 384, 88 S.Ct. 967, 971, 19 L.Ed.2d 1247 (1968). See United States v. Baxter, 492 F.2d 150, 172 (9th Cir.1973), cert. denied, 414 U.S. 801, 94 S.Ct. 16, 38 L.Ed.2d 38 (1973), 416 U.S. 940, 94 S.Ct. 1945, 40 L.Ed.2d 292 (1974).” Excerpt of Record at 95-96.
The district court’s findings were on a mixed question of law and fact. Mata II, 102 S.Ct. at 1306. As such, they are entitled to deference on our part. At the moment, we apply a “clearly erroneous” standard of review to a district court’s finding in a mixed question of law and fact. Maxwell v. Sumner, 673 F.2d 1031, 1035-36 & n. 5 (9th Cir.1982).
The majority does not address the standard of review of the district court opinion, and therefore does not explain why it has chosen not to defer to the district court. It should have done so.
The district court did not, of course, have the benefit of guidance from the Supreme Court on the importance of 28 U.S.C. § 2254(d), as did the majority in its previous and present opinions. The district court, therefore, did not explain whether it complied with section 2254(d) in evaluating the Fifth District Opinion, or whether it made a de novo examination of the record.
As I noted in my second dissent in this case, it might be sensible to remand this case to the district court to give Mata the opportunity there to rebut the Fifth District’s findings. 649 F.2d at 718 (Sneed, J., dissenting). As our three-fold consideration of Mata’s claims vividly illustrates, it is not in the interests of judicial economy to have a Court of Appeals review the facts of this type of petition when a district court is equally able to do so.
. The district court stated that “even if the testimony of Allen was tainted, the positive identification of petitioner by witness Childress and the strong one by Almengor makes [sic] the error harmless beyond a reasonable doubt. Chapman v. California, 386 U.S. 18 [87 S.Ct. 824, 17 L.Ed.2d 705] (1967).” Excerpt of Record at 96. It should be noted that the district court followed the same harmless error standard as the present majority. See Majority Opinion at 1255.
. Respondents summarize this evidence: “Childress testified that he was acquainted with petitioner [Mata] and co-defendant Gallegos prior to the crime, (RT 268) that he witnessed the crime (RT 275-278) and he made in-court identifications of Mata, Gallegos and Vargas as the assailants (RT 279). Childress testified that on October 20, 1972, the day after the incident, he reported to Lieutenant Mills what he had witnessed and verbally identified Mata and Gallegos by their known names, “Biggs” and “David” (RT 284 — 285). Childress testified that he then identified all three defendants’ photographs from the Rolodex of pictures of hun*1264dreds of inmates (RT 285-286). Lieutenant Mills testified that he isolated the photographs of Mata and Gallegos and Childress confirmed their identity and then again picked their photos from the Rolodex (RT 676-677). Childress testified that he never had any doubt about his identification of the three defendants (RT 286-287).” Respondent’s Return to the Order to Show Cause and Points and Authorities in Opposition to Petition for Writ of Habeas Corpus, Excerpt of Record at 42.
. “Carlos Ortega, the chief of the Mexican Mafia at Techachapi[,] testified that petitioner Mata and defendant Vargas, also members of the Mafia, contacted him and Mata told him that the victim was due to arrive at the prison, and they were ordered “to hit” the victim (RT 822-824). The next day, October 19, he and petitioner Mata and defendant Gallegos met, and they urged Ortega “to get down” meaning to arm himself with a knife and stab the intended victim (RT 82[8]). Mata and Gallegos departed, and defendant Vargas approached Ortega. Mata and Gallegos returned, Mata tried to give Ortega a knife which Ortega identified as one of the murder weapons, and Mata said he had another knife in his belt (RT 8[29]-831). Ortega refused, and Vargas called him a “cry baby” in Spanish (RT 832).
“Ortega and Gallegos were among several inmates transferred to Tracy after the incident. En route, Gallegos said, “My knife broke when I hit him. We all worked out on it.” (RT 83 [8]).” Id. at 48.
. “Joe Escobar testified that he is a member of the rival group, Nuestra Familia, the group to which the victim also belonged (RT 1154). Just before the stabbing took place, Escobar saw petitioner and his two co-defendants walking toward dorm 8 where the stabbing took place (RT 1175-1178). Escobar saw co-defendant Gallegos a short time later; his face was flushed and he appeared excited (RT 1178). That evening Escobar saw petitioner Mata who motioned to himself and smiled, indicating that he had “hit” the victim (RT 1183-1184).” Id.
. The majority also contends that Almengor’s description of Mata to the police “clearly was not detailed.” I fail to see how the majority reached this conclusion. Almengor described the height, build, weight, complexion, and clothes of Mata. R.T. at 419, 969-71. The description in Neil v. Biggers, characterized by the Court as “more than ordinarily thorough,” 409 U.S. at 200, 93 S.Ct. at 383, only added to Almengor’s list the age, skin texture, and voice of the suspect. Id. But even if the majority’s contention were true, the Neil v. Biggers test refers us to the “accuracy” of the prior description, not the amount of detail in it. Id. at 199, 93 S.Ct. at 382; United States v. Hanigan, 681 F.2d 1127, 1133 (9th Cir.1982). See Mata II, 102 S.Ct. at 1306 n. 6. Moreover, the inadequacy — or even absence — of a pre-trial description of the defendant by a witness in this context carries little weight in deciding whether a pre-trial identification is “reliable.” See United States v. Hammond, 666 F.2d 435, 440 (9th Cir.1982).
. The majority’s statement to the contrary, Majority Opinion at 1255, finds no support in the record. R.T. at 444 45.
. I wrote: “The prison setting of the crime for which appellant was convicted and the investigation thereafter by prison officials in my opinion dictate the use of photographic identification procedures. Apparently the majority insists as a matter of constitutional law that there be employed lineups of a large group of inmates, with each suspect, or more likely a substantial group of inmates, which would include the suspects, suitably equipped with counsel, in lieu of photographic identification procedures. See [611 F.2d at] p. 759 n. 1. To state the requirement reveals its impracticability. Moreover, it would impose heavy demands on the staff, strain employee relations, and expose the inmates to increased risks of bodily harm.
“The prison world is unique. It differs enormously even from the precinct station house and police headquarters. A “code of silence” strengthened by taboos against “ratting” and a pervasive fear of retaliation are characteristics of the prison social order. In this environment prison administrators and guards must function. Administrators are responsible for protecting prisoners in their custody and may be held liable for a failure to provide such protection .... Guards, directly responsible for prison order and security, jealously husband their stock of authority and seek to avoid any confrontation that will deplete that stock .... To require the type of lineup the majority envisions in this setting is to insist upon jeopardizing the security and safety of all as the price of securing the constitutional protection that the majority holds appellant is entitled. We should be reluctant to fashion constitutional doctrines whose price is so dear.” 611 F.2d at 760-61 (citations omitted).
. The array in Simmons was therefore smaller than the arrays used in the identification of Mata. Moreover, I doubt whether the present membership of the Court would agree with the Simmons Court’s dictum on photographic arrays. See, e.g., United States v. Ash, 413 U.S. 300, 319-21, 93 S.Ct. 2568, 2578-79, 37 L.Ed.2d 619 (1973).