dissenting:
I am unable to agree with the decision of the majority that Hoots was not deprived of his sixth amendment right to effective assistance of counsel. Thus, I would grant Hoots’ petition for a writ of habeas corpus. For that reason, I respectfully dissent.
I.
The majority has failed to properly apply the Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984) test to the facts of the instant case. Although I agree that attorney Weldon’s failure to solicit testimony from other eyewitnesses and impeach Roark with her criminal convictions satisfied the first prong of the Strickland test,1 I disagree with the conclusion that such unprofessional conduct did not meet the second prong.2 The facts clearly warrant a finding that Weldon’s deficient performance prejudiced Hoots’ defense.
In applying the second prong of the Strickland test, the majority did not consider the accumulative effect of Weldon’s two unprofessional errors. The majority evaluated the errors separately and determined that each one did not independently create a reasonable probability that but for the specific error, a different jury verdict would have occurred. See Majority Opinion at 1219-21.3 Strickland requires that this court consider together all of counsel’s deficiencies in determining “reasonable probability.”4 To do otherwise, as has the majority, results in treating the two attorney errors upon which Hoots relies for his petition for writ of habeas corpus, as two separate petitions.
II.
Under a proper application of Strickland’s second prong, there is a reasonable probability that, but for Weldon’s two unprofessional errors, the result of the proceeding would have been different. The jury would have found Hoots innocent.
The key witness for the prosecution was Karen Roark. As the majority stated, the state’s case against Hoots was completely dependent upon Roark’s testimony. She was the only witness who identified Hoots as the gunman. Thus, evidence contradicting and discrediting Roark’s testimony would have put doubts in the jurors’ minds as to the identity of the gunman. There is a reasonable probability that these doubts would have resulted in an acquittal of Hoots.5
Evidence contradicting Roark’s testimony could have been obtained through the testimony of the other three eyewitnesses to the robbery. Roark had testified that she recognized the gunman as being Hoots because he wore a sheer green scarf only across his face and thus she could discern his facial and hair features. She described his size, hair color, and affirmatively as*1224serted that he had no mustache. Roark also stated that she never lost eye contact with the gunman. This testimony would have been disputed by the other eyewitnesses.
Davis, Goins and Hall would have testified that the gunman and his features were not identifiable because of the covering over his face and head. Davis would have claimed the gunman had eye contact with her. Such corroborating evidence raises substantial doubt about Roark’s testimony regarding the face covering and her ability to see any characteristics of the gunman. In addition, a single eyewitness’s positive identification of a gunman is seriously questioned when three other eyewitnesses cannot detect even a single facial feature.
Evidence of the three eyewitnesses’ inability to make out the gunman’s facial characteristics would have refuted Roark’s positive assertion that the gunman did not have a mustache. Conflicting evidence as to the gunman's facial hair, coupled with the testimony disputing Roark’s positive identification, could have led to Hoots’ exoneration.
Substantial evidence at trial implicated another man, Darrell Shaw, as the gunman. He had the same body size, hair color, and clothing on the date of the robbery as the gunman at the Pizza Hut. Moreover, the undisputed evidence showed that Shaw and Hayes, the other Pizza Hut robber, were together shortly before and after the robbery and had borrowed the gun used in the crime. Shaw, however, had a mustache. This feature directly conflicted with Roark’s description of the gunman and thus implied Shaw’s innocence. Hoots, however, met Roark’s description (i.e., he was similar in size and hair color and had no mustache).
In finding Hoots to be the gunman, and not Shaw, the jury gave more weight to Roark’s undisputed testimony than to the other evidence implicating Shaw.6 If Davis, Goins, and Hall had testified that the gunman’s facial features were not discern-able, it is reasonably probable that the jury, in rendering its verdict, would have doubted Roark’s description and identification of the gunman, thereby relying more on the evidence against Shaw and Hoots’ alibi.
The majority contends that the testimony of Davis, Goins, and Hall would have been discounted because the lighting where Roark was positioned was better than the lighting where the other three witnesses were located. This distinction in lighting is not significant enough to presume the jury would have considered Roark’s testimony more reliable.
First, the lighting around the gunman, not the witnesses, is more important when discerning features through a covering. An effective counsel could have argued this point at trial. Moreover, the slight differences in lighting cannot extenuate the substantial discrepancies in the eywitnesses’ testimony. Roark stated she could see that the gunman had blond kinky-like hair, as he was only wearing a facial covering. The other three witnesses would have stated unequivocally that the gunman had a covering over his head. In addition, Roark stated that she never lost eye contact with the gunman. Davis, who was the same distance, but in a different direction, from the gunman as was Roark, would have testified that the gunman looked primarily at her (Davis) and only glanced occasionally at Roark.
These unexplained discrepancies would have required the jury to assess the credibility of Roark and the other three eyewitnesses. Considering the totality of the evidence and would-be evidence, it is probable that the jury would have given little credibility to Roark’s testimony. The corroborating testimony of three uninterested eyewitnesses appears to be more worthy of belief than the contrary testimony of a single witness. This is especially true when the single witness, like Roark, has *1225been convicted of criminal acts, which indicate her tendency to be dishonest and untruthful. As a result of this discrediting of the state’s sole identification evidence, it is reasonably probable that the jury would have found the evidence insufficient to prove beyond a reasonable doubt that Hoots was the gunman.
The majority does not consider Roark’s worthless check convictions as an implication of her testimonial untrustworthiness. I disagree. Intentionally cashing worthless checks is dishonest and deceitful. Indeed, this circuit has stated that convictions for cashing worthless checks under false pretenses or any other crimen falsi bears directly upon the witness’s propensity to testify honestly and truthfully. See United States v. Cunningham, 638 F.2d 696 (4th Cir.1981).
The majority also underestimates the probative value of using Roark’s criminal conviction to impeach her by speculating that the jury could have had a negative reaction to the impeachment. I do not believe the jury would have had such reaction. The majority’s speculation is based on the assumption that Roark’s identification testimony was credible and not contradicted by other evidence. Had Davis, Goins, and Hall testified, Roark’s testimony would have become suspect, thereby raising doubts as to Roark’s testimonial truthfulness. These doubts would have made the jurors more amenable to hearing impeaching statements against Roark.
In sum, Weldon’s unprofessional errors were devastating to Hoots’ defense. The testimony of Davis, Goins, and Hall would have contradicted the state’s sole identification witness. This would have led the jurors to doubt the accuracy of Roark’s description and identification of the gunman. The jury then probably would have given more weight to the other evidence, Hoots’ alibi and Shaw’s participation. These results, coupled with evidence (i.e., her criminal convictions) impeaching Roark’s credibility, would have culminated in the prosecution’s failure to meet its burden of proof. Thus, there is more than a reasonable probability that, but for the omission at trial of the testimony of Davis, Goins, and Hall and the proof of Roark’s past criminal convictions, the jury would have found Hoots innocent of being the gunman at the Pizza Hut on the night of July 24, 1979. Accordingly, I would reverse the district court and grant Hoots’ petition for a writ of habeas corpus.
. The first prong is whether counsel’s performance was deficient. Strickland, 466 U.S. at 686-88, 104 S.Ct. at 2064-65, 80 L.Ed.2d at 693.
. The second prong is whether "there is 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.” Id. at 694, 104 S.Ct. at 2068, 80 L.Ed.2d at 698.
. Hoots relied upon three errors committed by Weldon to illustrate the ineffective assistance: (1) failure to interview and use at trial the other eyewitnesses; (2) failure to impeach Roark; and (3) failure to introduce evidence on the third party commission of crime defense. The majority applied the reasonable probability test two times; once to error (1) and once to error (2). The test should have been applied only once, to errors (1) and (2) conjunctively. I do agree, however, with the majority’s opinion regarding error (3). See Majority Opinion at 1221-23.
. Indeed, in enunciating the second prong of the Strickland test, the court used the plural of “error.” See supra footnote 2.
. The state had to prove Hoots’ guilt beyond a reasonable doubt.
. In addition to the previously discussed evidence directly implicating Shaw, Hoots also presented four alibi witnesses who testified that Hoots was in bed sick in another town before and during the time of the robbery.