GILLERMAN, J.

In a secluded area of the outdoor Metropolitan Bay Transportation Authority Valley Road station in Milton, at about 7:15 A.M. on April 17, 1987, the victim was raped, robbed and beaten by a man who threatened her with a gun and then a knife. The victim identified the defendant as her assailant from a photographic array, in a police lineup and in court. On March 4, 1988, a jury returned verdicts of guilty on the charges of aggravated rape and armed robbery. On appeal the defendant claims that he was erroneously denied a peremptory challenge, that the prosecutor was allowed an improper challenge, and that certain testimony of an expert was erroneously admitted in evidence.

1. Jury Composition.

Certain background facts are pertinent. The victim was white and the defendant black; see Commonwealth v. Sanders, 383 Mass. 637, 640-641 (1981). During jury empanelment the prosecutor first challenged the only black person in the entire venire. The judge disallowed the challenge after hearing the prosecutor's reasons. The defendant then challenged eight women and one man when the panel consisted of twelve women and two men. The prosecutor objected, but the challenges were allowed, the judge noting the fact that the panel was overwhelmingly female. He said he would be more concerned if the panel were evenly divided between men and women and the defendant challenged all the women; "It's too early," he said, "to draw any inference of discrimination."

After replacement jurors were seated, the panel was seven men and seven women. The defendant challenged two white women. The prosecutor objected to these challenges on the grounds that both were young, female and white. The judge asked the defendant to explain the basis for his challenges. See Commonwealth v. Wood, 389 Mass. 552, 560-561 (1983). The defendant replied that he could not divulge his client's reasons. The judge said, "I can only assume that this is a discrete group that you have singled out and that you are exercising your peremptory challenges in violation of Soares."

There was more colloquy, and defendant's counsel, without divulging his client's reasons, disclosed his own: one was a secretary in the House of Representatives; the other juror came from Quincy, not far from where the crime took place. The judge found the explanations unsatisfactory, and he ruled that the defendant could exercise one, but not both, of the last two challenges. The defendant objected, and, as he was then obliged to do, made his choice between the two jurors.

Then followed a conference at the side bar where the judge explained the basis for his ruling:

"THE COURT: For the record, I would like to note that there are three young ladies of the same age group. Miss . . .[X] who is black was born in 1964. She was not challenged. Ms. . . . [Y] who is white was born in 1966, and she was challenged, and . . . [Miss Z] who is white was born in 1961. Those three appear to be the only three in that age group. The two whites were both challenged, and that was the reason for the Court's conclusion that a challenge of those through peremptory challenges was an abuse of the peremptory challenge and that the explanation given by counsel appeared to the Court to be a pretext and not a valid reason. That is the reason for the Court's action that the defense shall not be deprived of any peremptory challenges which he has just exercised."

The group identified by the judge consisted of three persons all of whom were young (under age thirty) and female. Within that group two were white and one was black. The group of three was defined by one characteristic — age — that could provide a permitted basis for a peremptory challenge, and one characteristic — gender — that was not a permitted basis. The judge ruled that when the defendant challenged only the two whites and not the entire group of three, he had impermissibly focused on race. The fact that the group of three was defined partly by age does not help the defendant. When the basis for the challenge, if only in part, is membership in a discrete group identified in art. 1 of the Declaration of Rights, the challenge is not permitted. Compare Commonwealth v. DiMatteo, 12 Mass. App. Ct. 547, 553 (1981). There was no abuse of discretion. Commonwealth v. Soares, 377 Mass. 461, 490, cert. denied, 444 U.S. 881 (1979) ("Although decisions of this nature [i.e., peremptory challenges] are always difficult, we are convinced that trial judges, given their extensive experience with jury empanelment, their knowledge of local conditions, and their familiarity with attorneys on both sides, will address these questions with the requisite sensitivity"); Commonwealth v. DiMatteo, supra, at 552 ("Sorting out whether a permissible or impermissible reason underlies a peremptory challenge is the function of the trial judge, and we do not substitute our judgment for his if there is support for it on the record"); Commonwealth v. Harris, 28 Mass. App. Ct. 724, 728 (1990). The fact that the judge permitted the defendant to exercise one of the two impermissible challenges is obviously a ruling of which he should not complain. Commonwealth v. Wood, 389 Mass. 552 (1983), upon which the defendant relies, is distinguishable. There the basis for the challenge was determined to be age, not gender.

After the panel was evenly divided between seven men and seven women, the prosecutor challenged five men. The judge said it "looks like a discriminatory exercise of challenges," and asked the prosecutor to explain. She did, and the judge concluded that there was a "possible argument," with regard to one juror, but that the reasons advanced for the other four were unacceptable. He ruled that the prosecutor could challenge the juror as to whom there was a "possible argument," and she could challenge one of the remaining four. The defendant objected on the ground that the judge had found the reasons advanced for all four not acceptable. The judge replied, "I appreciate what you are saying. I have already explained what the difference is. When there were five it was different. When it's two [of the five — one of whom is arguably satisfactory], you may have both. . . ."

There was no error. The judge first evaluated the reasons advanced by the prosecutor in the context of a divided panel with challenges to five of the seven men. It was well within the discretion of the judge to find that when the group is reduced to one, the same reason, previously advanced and rejected, may become tolerable. The judge was acting within his discretion. Commonwealth v. Thomas, 19 Mass. App. Ct. 1, 4 (1984) ("We note that the distinction . . . is between good faith and bad faith, not good and bad explanations").

2. Hair Analysis.

Hair samples taken from the defendant with his permission were submitted to the Commonwealth for examination. At the trial, Jay E. Godleski, an employee of the Department of Public Safety crime laboratory, qualified as an expert hair chemist. He testified that, based upon a microscopic examination and comparison of hair samples taken from the defendant and a hair fragment found on the victim's shirt, the recovered hair fragment, in his opinion, was consistent with the defendant's hair. The defendant properly makes no claim of error at this point. See Commonwealth v. Tarver, 369 Mass. 302, 310 (1975).

The defendant does challenge, however, the judge's admission of Godleski's subsequent testimony that, with regard to the recovered fragment, only one person in 4,500 would have hair with the same characteristics. "Where courts and commentators have been reluctant to admit statistical evidence, that reluctance has stemmed largely from the fact that the probabilities on which the evidence depended were based on mere speculation or were characterized in such a way as to mislead or confuse the jury." Commonwealth v. Gomes, 403 Mass. 258, 274 (1988). Statistical evidence, if relevant, may be admissible if the testimony is based on facts and not conjecture. At first, the judge was prepared to exclude the testimony because of the absence of the necessary factual foundation. However, additional testimony by Godleski established that the probability numbers he had used had been developed by the Royal Canadian Mounted Police in a 1974 study of 100 people who were representative of the Canadian population. Two black persons were included in the study group. That study, which involved some 800,000 hair comparisons, was, according to Godleski, generally accepted by the scientific community involved in the examination of hairs. The judge concluded that the testimony had a sound foundation, was not speculative, and therefore was admissible. We agree.

This is not a case where the witness used testimony regarding probability as a basis for his opinion that the defendant is the person whose identity is in question. See, e.g., Commonwealth v. Drayton, 386 Mass. 39, 49-50 (1982). Here Godleski never testified to more than that the defendant's hair was consistent with the recovered fragment of hair, and that the probability of its being consistent with the hair of another person was 1 in 4,500. "One hair cannot be used to identify an individual," Godleski concluded. The admissibility of the testimony was entirely within the discretion of the judge, and he exercised it wisely. See Commonwealth v. Garabedian, 399 Mass. 304, 310 (1987).

It is important to note here that the victim's testimony with regard to the identification of the defendant was exceptionally strong. The attack lasted about twenty to twenty-five minutes and the victim had more than ample time to see and remember the appearance of her assailant. Subsequently the victim identified the defendant in a photo array, at a lineup and in court. She testified she was one hundred percent certain that the defendant was the person who raped, robbed and beat her. Whatever slight infirmity may have attached to the Canadian study was not, in these circumstances, fatally prejudicial. See Commonwealth v. Dunn, 407 Mass. 798 (1990.)

Judgments affirmed.