FRIENDLY, Circuit Judge:
On this appeal from an order of the District Court for Connecticut denying a state prisoner’s petition for habeas corpus, we are confronted, as we recently were in
United States v. Reid,
517 F.2d 953, 965-67 (2 Cir. 1975), with a defendant’s claim that his constitutional rights were compromised by the prosecution’s display of his photograph singly to a witness for identification. Although the facts are much less favorable to the prosecution than in
Reid,
the State claims to be entitled to prevail under the Supreme Court’s latest identification decision,
Neil v. Biggers,
409 U.S. 188, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972).
I.
The district court was not asked to conduct an evidentiary hearing, and we take the facts, as it did, from the state trial record.
Trooper Glover of the Connecticut State Police, who had been assigned to the Hartford narcotics squad in an undercover capacity, went with an informant, Henry Brown, around 7:45 p. m. on the evening of May 5, 1970, to an apartment on the third floor of a building at 201 Westland Street.
He knocked at the door of the apartment. When it was opened, he observed a man standing in front of a woman and asked for “two things” of narcotics. The door was closed for a few moments and then was reopened. The man had two glassine envelopes containing a white powder, later determined to be heroin, for which Glover paid $20.
Glover was
within two feet of the seller and observed his face for two or three minutes; natural light was coming through a window at the hallway and Glover claimed to have had no difficulty in seeing.
When Glover left the building, he reported to a back-up officer outside, Detective D’Onofrio. He described the seller as being “a colored man, approximately five feet eleven inches tall, dark complexion, black hair, short Afro style, and having high cheekbones, and of heavy build. He was wearing at the time blue pants and a plaid shirt.” D’Onofrio went back to the police records division and obtained a photograph of petitioner Brathwaite. D’Onofrio testified that he selected this photograph because he recognized Brathwaite as the person described by Glover and that he had previously seen Brathwaite “several times, mostly in his vehicle.”
D’Onofrio took the photograph to the office of Glover’s squad. On May 7 Glover identified it as depicting the seller of the narcotics. For reasons not disclosed by the record, Brathwaite was not arrested until late July, 1970; the arrest took place in Mrs. Ramsey’s apartment at 201 Westland Street, see n. 2.
At the trial in January, 1971, Glover testified to the photographic identification and also made an in-court identification of Brathwaite, who was sitting at the defense counsel table. He had no doubt about the identifications. The state presented no other evidence to show that Brathwaite was the seller.
Brathwaite testified that at the time of the sale he was at home, suffering from a variety of ailments including a serious back condition which had kept him from going out for several days. On the following day, May 6, he went to a doctor’s office pursuant to a previously made appointment. According to him, Mrs. Ramsey was a friend of the family, who had driven his car when his back condition prevented him from doing so; she had called for him after he had had a myelogram at a hospital in July and also had driven him to her apartment on the day he was arrested there. Mrs. Brathwaite confirmed that her husband had been ill at home all day on May 5 and testified that Mrs. Ramsey had driven him to a doctor’s office on May 6. Dr. Vietzke testified that Brathwaite had been assigned to him as a clinic patient on April 15; that he had found a lack of sensation in Brathwaite’s legs which could (and ultimately was found to) indicate a disc involvement; that Brathwaite “moved like a man in great discomfort”; and that he referred Brathwaite to Dr. Owens, a neurosurgeon. Brathwaite’s May 6 appointment was at the neurosurgical clinic where he was seen by Dr. Owens.
The jury having returned a verdict of guilty, Brathwaite appealed his conviction to the Supreme Court of Connecticut which affirmed,
State v. Brathwaite,
164 Conn. 617, 325 A.2d 284 (1973). The portion of its opinion dealing with the identification issue is as follows, 164 Conn, at 619, 325 A.2d at 285:
The defendant claims that the court erred in permitting officer Glover to make an in-court identification of the defendant. The defendant asserts that the court should have determined whether evidence of Glover’s observance of the defendant’s photograph shortly after the sale was prejudicial before it allowed the in-court identification of the defendant. There was no objection or exception to the evidence when offered and this claim first appears in the defendant’s brief. The defendant has not shown that substantial injustice resulted from the admission of this evidence. Unless
substantial injustice is shown, a claim of error not made or passed on by the trial court will not be considered on appeal.
State v. Bausman,
162 Conn. 308, 315, 294 A.2d 312;
State v. Fredericks,
154 Conn. 68, 72, 221 A.2d 585.
This was followed by a petition for federal habeas and its denial by the district court.
II.
We must first consider the effect of the lack of objection to either the in-court or the photographic identification. As Judge Blumenfeld noted, this has two closely related aspects — the failure to exhaust state remedies and the effect of the state’s contemporaneous objection rule on the availability of federal habeas.
As we read the opinion of the district judge, he regarded the consideration of petitioner’s identification argument by the Supreme Court of Connecticut as meeting the exhaustion requirement of 28 U.S.C. § 2254(b) and (c). We are not so sure. It is, of course, true that plenary consideration of such an objection by a state appellate court meets the exhaustion requirement even though, under ordinary procedural rules, the court was not obliged to give this. However, our reading of its opinion leads us to believe that the Supreme Court of Connecticut considered the point on a more limited basis, namely, whether even if petitioner’s objections were sound, “substantial injustice resulted” from receipt of the identifications. It is doubtful whether such limited review would meet the exhaustion requirement if Connecticut provided a method for raising the federal claim in a collateral attack. The presentation to a state appellate court which obviates any need for resort to state collateral proceedings,
Brown
v.
Allen,
344 U.S. 443, 448-49 n.3 73 S.Ct. 397, 97 L.Ed. 469 (1953), is predicated on a disposition on the merits which would foreclose a successful collateral proceeding. We believe, however, that the Connecticut Supreme Court’s disposition of petitioner’s claim, while less than a full consideration on the merits, would be considered by the Connecticut courts to be enough to bar a subsequent collateral proceeding since “Connecticut’s rule [is] that [state] habeas corpus cannot serve as an appeal for questions which might have been raised for direct review,”
United States v. Menser,
247 F.Supp. 826, 829 (D.C.Conn. 1965);
Wojculewicz v. Cummings,
143 Conn. 624, 628, 124 A.2d 886, 890-91 (1956). If this be so, the exhaustion requirement of 28 U.S.C. § 2254 would pose no obstacle to petitioner since “§ 2254 is limited in its application to failure to exhaust state remedies still open to the habeas applicant at the time he files his application in federal court.” Fay v.
Noia,
372 U.S. 391, 435, 83 S.Ct. 822, 847, 9 L.Ed.2d 837 (1963).
However, it is unnecessary to decide this since the respondent has not here raised a claim of failure to exhaust state remedies, and the requirement is not jurisdictional but merely a principle of comity,
id.
at 434-35, 83 S.Ct. 822. For the same reason, namely, the State’s failure to raise the point here, we need not consider whether Fay v.
Noia, supra,
372 U.S. at 426-34, 438-40, 83 S.Ct. 822, would be read today as relieving Brathwaite, in federal habeas, of his counsel’s procedural default, even though, for the reason indicated above, we doubt whether the Connecticut Supreme Court’s limited consideration of Brathwaite’s objections would bring the case within the principle of
Warden v. Hayden,
387 U.S. 294, 297 n. 3, 87 S.Ct. 1642, 18 L.Ed.2d 782 (1967), that a state , may not assert “deliberate by-pass” when a petitioner’s constitutional claims have in fact been passed upon by a state court despite breach of a contemporaneous objection rule.
III.
Respondent concedes that exhibition of the single photograph of Brathwaite to Glover was “impermissibly suggestive” within many decisions of this court. See, e.
g., United States ex rel. Gonzalez v. Zelker,
477 F.2d 797, 801 (2 Cir. 1973),
cert. denied,
414 U.S. 924, 94
S.Ct. 254, 38 L.Ed.2d 158 (1974);
United States ex rel. John v. Casscles,
489 F.2d 20, 24 (2 Cir. 1973),
cert. denied,
416 U.S. 959, 94 S.Ct. 1976, 40 L.Ed.2d 310 (1975);
United States v. Reid, supra.
Not only was it impermissibly suggestive
but, despite the unexplained statement to the contrary in respondent’s brief, it was unnecessarily so. There was no emergency that prevented D’Onofrio from assembling a suitable array of photographs from the many fitting Glover’s description that must have been available in the files of the Hartford Police Department. In fact there was so little urgency that two days passed before
Glover
identified Brathwaite’s photo and D’Onofrio waited for
three
months before making Brathwaite’s arrest. D’Onofrio selected the single photograph not because of any difficulty in securing more but because, without having witnessed the event, he was certain Brathwaite had made the sale, see fn. 5.
Prior to
Neil v. Biggers, supra,
this alone would have permitted a speedy resolution of the case. This court and others had held that, except in cases of harmless error, a conviction secured as the result of admitting an identification obtained by impermissibly suggestive and unnecessary measures could not stand, see, e.
g., United States v. Fernandez,
456 F.2d 638, 641-42 (2 Cir. 1972);
Kimbrough v. Cox,
444 F.2d 8 (4 Cir. 1971);
United States v. Fowler,
439 F.2d 133 (9 Cir. 1971);
Mason v. United States,
134 U.S.App.D.C. 280, 414 F.2d 1176 (1969), although, following the lead of
United States v. Wade,
388 U.S. 218, 239-40, 87 S.Ct. 1926, 18 L.Ed.2d 1149 (1967), we allowed the prosecution to bypass the tainted identification and introduce subsequent identifications or have the witness make an in-court identification
if satisfied that these stemmed from the original observation of the defendant rather than the tainted identification.
United States ex rel. Phipps v. Follette,
428 F.2d 912 (2 Cir. 1970),
cert. denied,
400 U.S. 908, 91 S.Ct. 151, 27 L.Ed.2d 146 (1970);
United States ex rel. Gonzalez v. Zelker, supra,
477 F.2d at 801-05. Here there would have been no occasion to consider whether Glover’s in-court identification rested on his original observation (except for the bearing of this on a new trial) since the admission of the photographic identification would have been fatal constitutional error.
A passage in
Neil v. Biggers, supra,
409 U.S. at 198, 93 S.Ct. at 381, calls this analysis into some question. After saying that “the primary evil to be avoided is ‘a very substantial likelihood of irreparable misidentification,’ ” the inner quote being from
Simmons v. United States, supra,
390 U.S. at 384, 88 S.Ct. at 971, Mr. Justice Powell continued:
While the phrase was coined as a standard for determining whether an in-court identification would be admissible in the wake of a suggestive out-of-court identification, with the deletion of “irreparable” it serves equally well as a standard for the admissibility
of testimony concerning the out-of-court identification itself. It is the likelihood of misidentification which violates a defendant’s right to due process, and it is this which was the basis of the exclusion of evidence in
Foster.
Suggestive confrontations are disapproved because they increase the likelihood of misidentification, and unnecessarily suggestive ones are condemned for the further reason that the increased chance of misidentification is gratuitous. But as
Stovall
makes clear, the admission of evidence of a showup without more does not violate due process. (Footnote omitted.)
If this stood alone, it would strongly support a conclusion that the Court intended the “very substantial likelihood of misidentification” test
to apply to show-up or photographic identifications that were impermissibly and unnecessarily suggestive as well as to later out-of-court or in-court identifications.
However, Mr. Justice Powell also said, 409 U.S. at 198-99, 93 S.Ct. at 382:
What is less clear from our cases is whether, as intimated by the District Court, unnecessary suggestiveness alone requires the exclusion of evidence. . . . The purpose of a strict rule barring evidence of unnecessarily suggestive confrontations would be to deter the police from using a less reliable procedure where a more reliable one may be available and would not be based on the assumption that in every instance the admission of evidence of such a confrontation offends due process.
Clemons
v.
United States,
133 U.S.App.D.C. 27, 48, 408 F.2d 1230, 1251 (1968) (Leventhal,
J.,
concurring); cf.
Gilbert v. California,
388 U.S. 263, 273, 87 S.Ct. 1951, 18 L.Ed.2d 1178 (1967);
Mapp v. Ohio,
367 U.S. 643, 81 S.Ct. 1684, 6 L.Ed.2d 1081 (1961). Such a rule would have no place in the present case, since both the confrontation and the trial preceded
Stovall v. Denno, supra,
when we first gave notice that the suggestiveness of confrontation procedures was anything other than a matter to be argued to the jury.
Although the commentators differ concerning the meaning of this passage,
we think that, at minimum, it preserves, in cases where both the confrontation and the trial were subsequent to
Stovall,
the principle requiring the exclusion of identifications resulting from “unnecessarily suggestive confrontation.” This construction is powerfully supported by the citations in the last quotation. The passage in Judge Leventhal’s concurring opinion in
Clemons v. United States,
408 F.2d at 1251, drew a sharp distinction between “post-Wade trials” and a case where “a trial, as well as the identification, has taken place prior to Wade and
Stovall.”
The
Clemons
major
ity had also said that the exclusionary rule “would appear to be applicable with respect to prosecution evidence of post
Stovall
pre-trial identifications found by the court to be violative of due process,” 408 F.2d 1236-37 (McGowan,
J.);
see also the characterization of the majority’s position in Judge Wright’s concurring and dissenting opinion, 408 F.2d at 1252-53. Equally clearly the citations to
Gilbert
and
Mapp
point to a principle of exclusion. It is true, as the commentators have pointed out, that there is some novelty in applying the non-retroactivity principle to this portion of the
Stovall
opinion, 388 U.S. at 301-02, 87 S.Ct. 1967, which reads on its face as if it were to have the usual application
as distinguished from the solely prospective authority decreed for the
Wade
and
Gilbert
decisions. But experience may have convinced the Court that retroactive application of this portion of
Stovall
also would have too drastic an effect on law enforcement; it was, as said in
Neil v. Biggers, supra,
409 U.S. at 199, 93 S.Ct. at 382, the first occasion on which the Court “gave notice that the suggestiveness of confrontation procedures was anything other than a matter to be argued to the jury.”
This interpretation of
Neil v. Biggers
—that it qualified the
Stovall
standard only with respect to
pre-Stovall
cases— has been adopted expressly by the Fourth Circuit,
Smith v. Coiner,
473 F.2d 877, 880-81 (1973),
cert. den. sub nom., Wallace v. Smith,
414 U.S. 1115, 94 S.Ct. 848, 38 L.Ed.2d 743 (1973), and seemingly by the Fifth Circuit,
Rudd v. State of Florida, 477
F.2d 805, 809 (1973), and the Sixth Circuit,
Workman v. Cardwell,
471 F.2d 909, 910 (1972),
cert. denied,
412 U.S. 932, 93 S.Ct. 2748, 37 L.Ed.2d 161 (1973). Cf.
Souza v. Howard,
488 F.2d 462, 465 (1 Cir. 1973),
cert. denied,
417 U.S. 933, 94 S.Ct. 2646, 41 L.Ed.2d 237 (1974). We do not consider that
United States v. Evans,
484 F.2d 1178, 1186 n. 8 (1973), committed us to a contrary position. In that case the prosecution offered only an in-court identification and the tainted previous photographic identification was introduced by defense counsel in an effort to impeach the direct testimony of the identifying prosecution witnesses. See
United States v. Evans,
72 Crim. 910 Transcript at 118, 136. We read the footnote as meaning only that the court rejected, in light of
Neil v. Biggers,
a
per se
exclusionary rule for in-court identifications because the prosecution has used a less reliable method for obtaining a prior identification when a more reliable one was available.
On the other hand, the treatment of the problem in
United States v. Boston,
508 F.2d 1171 (2 Cir. 1974), is consistent with our analysis.
These conclusions, based upon a textual analysis of
Neil v. Biggers
and the subsequent decisions of the courts of appeals, are ■ reinforced when the case is placed in its setting. A good starting place is Mr. Justice Brennan’s observation in
United States v. Wade, supra,
388 U.S. at 228, 87 S.Ct. at 1933:
The vagaries of eyewitness identification are well-known; the annals of criminal law are rife with instances of mistaken identification
and the large body of literature cited in footnotes 6 and 7, 388 U.S. at 228-29, 87 S.Ct. 1926 to support this. The
Wade
and
Gilbert
decisions represented one way of responding with respect to lineups; this was supplemented by the ruling in
Stovall
that due process was violated by receiving an identification arising from a confrontation which is “unnecessarily suggestive and conducive to irreparably mistaken identification,” 388 U.S. at 301-02, 87 S.Ct. at 1972. The subsequent statement in
Stovall
that “a
claimed violation of due process of law in the conduct of a confrontation depends on the totality of the circumstances surrounding it” was not a blanket invitation to use suggestive confrontations whenever a witness might reasonably be expected to be able to resist the suggestion. It was directed at the particular need for the confrontation there at issue — that the immediate showing of Stovall to the hospitalized victim, Mrs. Behrendt, was imperative
—not to the fact that the hospital identification was almost certainly correct. Significantly the Court did not refer to the large amount of circumstantial evidence, see 355 F.2d at 733-34, which made it almost certain that there was no misidentification.
The Court next encountered the problem in
Simmons v. United States, supra,
390 U.S. 377, 88 S.Ct. 967, 19 L.Ed.2d 1247.
In contrast to
Stovall,
where the witness had testified to the show-up identification as well as making an in-court identification, the witnesses in
Simmons
had done only the latter. It was in this context that Mr. Justice Harlan said, 390 U.S. at 384, 88 S.Ct. at 971:
[W]e hold that each case must be considered on its own facts, and that convictions based on eyewitness identification at trial following a pretrial identification by photograph will be set aside on that ground only if the photograph identification procedure was so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification.
We do not regard this language as departing from the brief statement in
Stovall.
Indeed Justice Harlan said his standard accorded with the Court’s “resolution of a similar issue” in
Stovall,
390 U.S. at 384, 88 S.Ct. 967. The language was different because the issue was different.
Stovall
held that where the police or a prosecutor had engaged in indefensible identification procedures, they must be deprived of the immediate fruits.
Simmons
held, consistently with the Court’s approach in
Wade
(which in turn had relied, 388 U.S. at 241, 87 S.Ct. 1926, on the doctrine with respect to fruits of a primary illegality announced in
Wong Sun
v.
United States,
371 U.S. 471, 488, 83 S.Ct. 407, 9 L.Ed.2d 441 (1963)), that such an error should not always deprive the Government of testimony of later identifications that might be essential to a conviction; it should do so only when the error was irreparable. See
United States ex rel. Phipps v. Follette, supra,
428 F.2d at 914 n. 2. In
Foster v. California, supra,
394 U.S. 440, 89 S.Ct. 1127, 22 L.Ed.2d 402, which, like
Stovall,
involved both testimony as to an improper lineup identification and an in-court identification, the Court returned to the
Stovall
formulation, 394 U.S. at 442, 89 S.Ct. 1127, and added that “it is the teaching of
Wade, Gilbert
and
Stovall, supra,
that in some cases
the procedure
leading to an eyewitness identification may be so defective as to make the identification constitutionally inadmissible as a matter of law,” 394 U.S. at 442-43 n. 2, 89 S.Ct. at 1128 (emphasis supplied).
As cases in the wake of
Stovall
and
Simmons
began to reach them, the courts of appeals tended to use the
Simmons
formulation. Stress was placed on such factors as the witness’ opportunity and incentive for observation at the time of the crime, the accuracy of descriptions furnished before the suggestive identification, the witness’ level of certainty, the lengths of time between the crime and identification and the trial, and even
the existence of other evidence tending to show that the identification was not mistaken. See, e.
g., United States ex rel. Rutherford v. Deegan,
406 F.2d 217 (2 Cir.),
cert. denied,
395 U.S. 983, 89 S.Ct. 2145, 23 L.Ed.2d 771 (1969);
United States ex rel. Phipps v. Follette, supra,
428 F.2d 912;
United States ex rel. Frasier v. Henderson,
464 F.2d 260 (2 Cir. 1972). But in the cases cited and at least most of the others the prosecution, following the lead of
Wade,
relied only on a subsequent, usually in-eourt identification,
and the language from
Simmons
was treated as a sort of combination of the
Wade
independent source principle with a caution — perhaps due to a realization that the effect of the prior identification could never be wholly eradicated — that this might not suffice to avoid reversal unless the court was fairly assured that no harm was being done. The prosecution generally did not even assert that a suggestive identification made without an imperative necessity, such as existed in
Stovall
or in the immediate confrontation cases such as
Bates v. United States,
132 U.S.App.D.C. 36, 405 F.2d 1104 (1968) (Burger,
J.),
and
Russell v. United States,
133 U.S.App.D.C. 77, 408 F.2d 1280,
cert. denied,
395 U.S. 928, 89 S.Ct. 1786, 23 L.Ed.2d 245 (1969), could themselves be received in evidence.
In light of this survey and our analysis of the opinion itself, we do not think
Neil v. Biggers
intended to change the rule previously applied, except to subject to the more lenient
Simmons
test impermissibly and unnecessarily suggestive
pre-Stovall
identifications. For such
post-Stovall
identifications the rule remains as
Stovall
and
Simmons
left it. Evidence of an identification unnecessarily obtained by impermissibly suggestive means must be excluded under
Stovall,
and the more lenient
Simmons
language and the criteria worked out under it apply only to subsequent identifications with the prosecution having the burden of proving that the precautionary conditions of
Simmons
have been met. No rules less stringent than these can force police administrators and prosecutors to adopt procedures that will give fair assurance against the awful risks of misidentification.
IY.
Even if we should be wrong in all this and
Neil v. Biggers
was intended to apply the
Simmons
test to post-Stova
11
show-ups or photographic displays that were impermissibly and unnecessarily suggestive, as well as to in-court identifications following upon them, the writ should issue here. Although Glover testified that the hallway was well lit by sunlight, we can take judicial notice that on May 5, 1970 sunset at Hartford took place at 7:53 p. m. While Glover was a “trained observer,” see
United States v. Reid, supra,
517 F.2d at 966, he was in a very different position from Agent Shea in that case. Shea had been the victim of a brutal assault; Glover was acting as an undercover agent — whose business was to cause arrests to be made — and his description of the suspect, while apparently covering Brathwaite, could have applied to hundreds of Hartford black males. The certainty of identification is far less persuasive when the expressions come from the lips of an undercover agent, especially under the circumstances developed in notes 2 and 3, than when they are the words of an ordinary citizen, whether a bystander or a victim. The in-court identification has little meaning; Brathwaite was at the counsel table and Glover must have made dozens of identifications in the eight months be
tween the narcotics sale and the trial. Perhaps the strongest bit of evidence to strengthen the identification was Brathwaite’s arrest in the very apartment where the sale was made. But Brathwaite offered an explanation of this which was not implausible, although evidently not credited by the jury, and the long and unexplained delay in his arrest is troubling. Assuming, which we do not believe, that
Simmons
states the appropriate test for both of Glover’s identifications, we hold that both were inadmissible. The danger that Brathwaite was convicted because he was a man whom Detective D’Onofrio had previously observed near the scene of the crime, thought to be a likely offender, and arrested when he was known to be in Mrs. Ramsey’s apartment, rather than because Glover really remembered him as the seller, is too great to let this conviction stand.
The judgment is reversed with instructions to issue the writ unless Connecticut gives notice of a desire to retry Brathwaite within twenty days after issuance of the mandate and the retrial occurs within such reasonable, period thereafter as the district judge may fix.