The appellant was tried and convicted upon two counts of robbery. In this court, he makes two contentions. The first is that his defense of alibi was established beyond a reasonable doubt. The second is that the identification of the appellant by two victims at a police lineup was rendered questionable by the fact that one man who had been selected for the lineup was dismissed after the appellant’s counsel had approved him.
According to the state’s evidence, two motels located in *414the same area in Seattle were robbed, one on November 30, 1969, and the other on December 15, 1969. At a police lineup conducted on December 29, 1969, the appellant was identified by the persons from whom the money had been taken on both occasions.
When the first of these robberies occurred, a young man was on duty at the motel lobby and was having sandwiches with a young woman friend in an adjoining anteroom. The occupants of this room were able to see anyone who entered the lobby by looking at a convex mirror which was hung behind the registration desk. The robber on that occasion entered the motel lobby alone. The man on duty went into the lobby to attend to his request for a room. While the two were facing each other, the robber took a gun from his pocket and pointed it at the man on duty. He demanded and received all of the money which his victim could find. He then went into the anteroom. Seizing the young woman by the hair with his right hand and holding the gun to her head with his left hand, he ordered her to turn around so that she would not be able to see him. He threatened to shoot her if the motel employee did not produce more money. The employee searched the desk drawers but was finally able to convince the robber that there was no more money and he left.
These victims described the robber to the police shortly after the event. The employee identified the appellant at the police lineup but the young woman was not able to recognize him, either at the lineup or at the trial. She said that she could not observe the robber’s features clearly in the convex mirror and that she had seen his face for only a few seconds. However, she described the general appearance, manner and speech of the robber, and this description fitted the appellant. The appellant was left-handed.
There was only one witness to the second robbery, the man in charge of the motel, who described how two men, at least one of whom had a gun, had entered the office of the motel and forced him to give them all of the money in the *415cash register. He. identified the appellant as one of the men at the lineup conducted 2 weeks later.
The conviction rested upon the testimony of these victims, since there was no circumstantial or other corroborating evidence offered. The appellant does not deny that he fits the description of the robber given by the victims to the police. Furthermore, he does not suggest that a verdict may not be based solely upon eyewitness identifications.1 We have sustained such a verdict in State v. Lynch, 176 Wash. 349, 29 P.2d 393 (1934). He does suggest, however, that the identifications were discredited by the fact that the lineup was conducted without the presence of a man who had been selected to appear in it and who, in the opinion of the appellant’s counsel, bore the closest resemblance to the appellant of any member of the lineup.
No suggestion is advanced that the lineup contained an insufficient number of persons to constitute a fair lineup or that it had any of the suggestive features condemned by the United States Supreme Court in United States v. Wade, *416388 U.S. 218, 18 L. Ed. 2d 1149, 87 S. Ct. 1926 (1967). The appellant cites no authority for the proposition that he is entitled to select the persons who shall appear in a lineup.
Assuming that the removal of one person from the lineup did weaken the quality of the identification, this is a matter which would affect the weight to be attached by the jury to the testimony of the witnesses and would not render that testimony inadmissible. The appellant’s counsel put before the jury the fact that a man was selected and then removed from the lineup, and we may assume that the jury gave it full consideration. The verdict indicates that the jury was not convinced that the identifications were discredited.
The remaining contention of the appellant is that his defense of alibi was established as to each crime. He testified that he was elsewhere at the time each of these crimes was committed; and he was corroborated by an additional witness in each case. These two witnesses were friends of the appellant. One said that he had been with the appellant on a fishing trip at the time of the first robbery, and the other said that she had seen him at a party in Pasco on the night of the second robbery. We have read the record and must conclude that the evidence on the question of alibi was not as “clear, cogent and convincing” as the appellant maintains. It is true it was uncontradicted except for the testimony of the victims of the robberies, who said the appellant was the person who robbed them. Assuming the appellant is correct that, in evaluating the alibi evidence, we must disregard the contradictory evidence of the eyewitnesses, we do not find the testimony of the appellant and his witnesses to be as free of inconsistency and as coherent and convincing as he insists that it is. In any event, we cannot say upon the record, as a matter of law, that the defense of alibi was established.
' The evidence being in conflict, it was for the jury to decide which witnesses' to believe. There is in the record of this case substantial evidence upon which the jury could *417find the appellant guilty beyond a reasonable doubt. This is the evidence which it chose to believe.
The judgment is affirmed.
Hamilton, C.J., Hale, Sharp, and Wright, JJ., concur.
Eyewitness identifications are by no means infallible. See E. Borchard, Convicting the Innocent (Errors of Criminal Justice) (1961) and P. Wall, Eye-Witness Identification in Criminal Cases (1965). In 29 of the 65 cases of erroneous convictions reported by Borchard, mistaken identification was practically alone responsible for the injustice. In 8 of those cases the wrongfully accused person bore not the slightest resemblance to the real culprit and in 12 the resemblance was not at all close. In only two cases was the resemblance striking. In 1 of the cases cited, the accused, although later proven innocent when the actual offender was apprehended, was identified by no less than 17 witnesses.
Wall concludes from his study that the problem of mistaken eyewitness identification is the most serious obstacle to the administration of criminal justice. The author recognizes that the dangers of mistaken identification can be greatly reduced if the witness identifies the accused in a properly constituted and conducted police lineup at a time when the events are fresh in the memory of the witness. (See State v. Simmons, 63 Wn.2d 17, 385 P.2d 389 (1963), recognizing that an extrajudicial identification at a time when the events are fresh in the mind of the witness is more likely to be reliable evidence than a courtroom identification.) He suggests a requirement of corroboration where such a prior identification has not been made and further suggests that the jury should be cautioned about the dangers of identification evidence. Some additional precautions of a more innovative nature are suggested in his concluding chapter.