After a jury trial, the defendant was convicted of two counts of third-degree criminal sexual conduct, MCL 750.520d(l); MSA 28.788(4)(1), and one count of breaking and entering with intent to commit criminal sexual conduct, MCL 750.110; MSA 28.305. Defendant was sentenced to concurrent prison terms of 5 to 15 years for each of the convictions and appeals as of right.
Defendant’s first claim on appeal is that the trial court’s failure sua sponte to give a cautionary instruction on tracking dog evidence was reversible error.
In People v Perryman, 89 Mich App 516, 524; 280 NW2d 579 (1979), this Court held that a trial court has a duty, even absent a request by counsel, to inform the jury that tracking dog evidence must be considered with caution, is of slight probative value, and cannot support a conviction in the absence of other direct evidence of guilt. However, in People v McRaft, 102 Mich App 204, 210-211; 301 NW2d 852 (1980), lv den 411 Mich 987 (1981), the Court found that the trial court’s failure sua sponte to give a cautionary instruction on tracking dog evidence was harmless error. The Court stated that the failure to give the cautionary instruction was not so offensive to the maintenance of a sound judicial system that it could never be regarded as harmless, that the prosecutor did not deliberately inject the error into the proceedings, that the error did not deny the defendant his fundamental rights, and that the error was not particularly inflammatory or persuasive. Further, the Court found that the error was harmless beyond a rea*215sonable doubt because of the overwhelming evidence against the defendant.
In the present case, we find that any error on the part of the trial court in failing to give the cautionary instruction was harmless. Defense counsel did not object to the admissibility of the evidence and did not request jury instructions regarding the evidence. Although the prosecutor commented on the evidence, his comments were not inflammatory. Further, defense counsel emphasized the inconclusive nature and deficiencies of the evidence in his closing argument. In addition, the evidence presented identifying the defendant as the assailant was overwhelming.
When the police arrived shortly after the assault, the complainant gave them a description of the man who had assaulted her and who she believed lived in the apartment complex. She described him as 5'7" or 5'8" tall, skinny, with olive colored skin, moustache, dark brown feathered hair, and "whisker holes” in his face. Although the complainant had observed her assailant for only a short time during the assault, she was able to give a detailed description because the man placed his face alongside hers and spoke to her continuously throughout the assault. The man told her that he was in a hurry and lived nearby. She noted that she had seen him before around the apartment building and at a barbecue. They occasionally exchanged "hellos” when she saw him near the shed in back of the apartment, but they had never met or talked. She believed that her assailant had seen her before because he told her that he always thought she was pretty.
She also noted at the time of the attack that the assailant was wearing a leather jacket and brown fur-lined gloves which were stretched out at the *216knuckles. The jacket had ribbed cuffs and epaulets. At trial, she identified defendant’s jacket and gloves as the same ones worn by the assailant. The assailant also wore a dark blue or black knit winter hat.
The complainant further testified that the man had a low, slow distinctive voice. She later heard the same voice over an extension phone at the state police post when defendant telephoned to speak to one of the troopers about the case. After listening to the defendant speak in the hallways outside the courtroom, she confirmed that his voice was the same as her assailant’s.
The day after the attack, the complainant and her sister returned to the apartment to pick up her car. As they were turning onto the service drive, the complainant observed the defendant driving a car. She immediately contacted the police. She also saw him later working near the garbage dumpster behind the apartment complex. The complainant was certain that the defendant was the man who raped her. Thus, based upon the overwhelming evidence identifying the defendant as the assailant, we find no reversible error.
Defendant’s next claim is that the trial court erred in admitting blood type evidence.
At the trial, the testimony revealed that the complainant’s assailant was a type O secretor, that 35% of the population have this blood type, and that the defendant is a type O secretor. Defense counsel objected to the admissibility of the evidence and argued that the evidence was more prejudicial than probative. Therefore, this issue has been preserved for our review.
This Court is split on whether blood type evidence is admissible for inclusory purposes. In People v Sturdivant, 91 Mich App 128, 131; 283 NW2d *217669 (1979), lv den 407 Mich 933 (1979), the Court ruled that it was error to admit blood type evidence which serves "to include the defendant in a class of possible assailants * * * thereby increasing] the probability of the defendant’s guilt without connecting him, in any way, to the charged offense”.
On the other hand, in People v Horton, 99 Mich App 40, 50-51; 297 NW2d 857 (1980), vacated and remanded on other grounds 410 Mich 865 (1980), the Court held that such evidence is admissible in accordance with the rules for the admission of other physical evidence. The Court further stated that, as the population group connected with a crime grows larger, the probative force of the evidence decreases but that the weight to be given to the evidence is for the jury’s determination. Several panels of this Court have followed the rule announced in Horton. People v Eaton, 114 Mich App 330, 336; 319 NW2d 344 (1982); People v Camon, 110 Mich App 474, 480; 313 NW2d 322 (1981), lv den 414 Mich 859 (1982); People v Young, 106 Mich App 323, 331; 308 NW2d 194 (1981), lv gtd 414 Mich 865 (1982).
The author of Sturdivant, Judge Kelly, subsequently modified the position he took in Sturdivant in his concurring opinion in People v White, 102 Mich App 156, 162-167; 301 NW2d 837 (1980). Judge Kelly stated that blood type evidence for inclusory purposes was less scientific than polygraphs and devoid of the identifiable characteristics of fingerprint, fingernail, hair and voice identification evidence. Further, the probative value of a narrow percentage classification would outweigh the potential for prejudice and such evidence would, therefore, be admissible. Judge Kelly also found that, in a criminal case which requires proof *218beyond a reasonable doubt, a conviction should not depend upon the prosecutor’s ability to place the defendant within a group of 20% of all males as in Sturdivant and Horton. In such large groupings, the benefit of including the defendant in the group is clearly outweighed by the potential for adverse influence on the jury. Judge Kelly opined that future cases involving smaller groupings or test results which more accurately point to a particular defendant would have to be decided on a case-by-case basis. The determination of admissibility should be left to the discretion of the trial court. Also, see People v Nixon, 114 Mich App 233, 239; 318 NW2d 655 (1982).
We find that the rule expressed by Judge Kelly is the better view and, therefore, find that the blood type evidence in the present case should have been excluded. The evidence placed the defendant in a class consisting of over one-third of the population and had little probative value, but the potential for prejudicing the jury was great.
However, we further find that any error in admitting the evidence in this case was harmless. Generally, harmless error has been found where the victim has positively identified the defendant and had a sufficient opportunity to observe him at the time of the attack. People v Baker, 114 Mich App 524, 530; 319 NW2d 597 (1982); People v Camon, supra, p 480; People v Sommerville, 100 Mich App 470; 299 NW2d 387 (1980); People v Sturdivant, supra, p 134. Although the complainant observed her assailant for only a few moments before he turned her around, she was able to see the side of his face and his jacket and gloves and was able to give a description to the police. She believed she had previously seen her assailant around the apartment building. She positively *219identified the defendant on the day after the assault and identified his clothing.
The defendant’s third claim of error is that the complainant’s voice identification of the defendant was the result of an impermissibly suggestive procedure which denied the defendant due process and violated his right to counsel. Police Officer Payne testified that four days after the assault on the complainant the defendant telephoned the officer at the police station and inquired about the investigation. Coincidentally, the complainant was present at the station. Officer Payne immediately directed her to pick up an extension phone. The defendant identified himself by name. After listening for a few moments, the complainant became very upset and indicated that the voice was that of her assailant.
Defendant failed to object to the testimony elicited from both the complainant and Officer Payne concerning the voice identification at the police station. Failure to object to testimony concerning an allegedly suggestive pretrial identification precludes appellate review absent manifest injustice. People v Moss, 397 Mich 69; 243 NW2d 254 (1976); People v King, 107 Mich App 208, 210; 309 NW2d 207 (1981), rev’d on other grounds 413 Mich 939 (1982). We find no manifest injustice in this case. Although the procedure may have been impermissibly suggestive, the complainant’s other identification testimony was substantial. Further, the complainant’s subsequent in-court voice identification of the defendant had a sufficient independent basis. The complainant had previously seen the defendant near the apartment and had exchanged greetings with him. Although she saw her assailant only momentarily during the attack, the man talked to her constantly. Shortly after the attack, *220she described the man’s voice as low and distinctive. There was no evidence that the complainant could not recall how the voice sounded.
Finally, the defendant claims that the cumulative effect of the above three alleged errors denied the defendant a fair trial since all three alleged errors related to the issue of identification. We disagree. The defendant is not entitled to a perfect trial, only a fair one.
Affirmed.
M. J. Kelly, J., concurred.