In this criminal appeal, the principal issue is the weight to be assigned to potentially exculpatory blood typing tests and deoxyribonucleic acid (DNA) profiling evidence. The state charged the defendant by substitute information with kidnapping in the first degree in violation of General Statutes § 53a-92 (a) (2) (A) and two counts of sexual assault in the first degree in violation of General Statutes § 53a-70 (a).1 After a jury trial finding the defendant *266guilty as charged, he moved for a new trial and for further discovery to undertake blood testing and DNA profiling of additional vaginal swabs and smears from the victim. Denying both motions, the trial court rendered a judgment sentencing the defendant to a total effective term of twenty-five years, suspended after twenty-three years, and three years probation. The defendant appealed to the Appellate Court, and we transferred his appeal to this court pursuant to Practice Book § 4023. We remand the case to the trial court for reconsideration of these two motions.
There is no dispute about the following facts. While walking on a dark street in Bristol, on November 30, 1987, at approximately 5:20 p.m., the victim was pushed off the sidewalk by a man who knocked her down and subsequently forced her to enter a vehicle in a nearby parking lot. The man drove the car for ten to fifteen minutes before parking it on or near a dirt road, whereupon he forced the victim to engage in oral and vaginal intercourse. She did not believe that the man ejaculated during the oral sex, and could not tell whether he ejaculated during the vaginal sex. The man then drove her to an area of Bristol with which she was unfamiliar, told her not to tell anyone what had happened or he would kill her and her son, and let her out of the car. He told her in which direction she would find Route 6, and drove away.
The defendant has consistently denied being the man who attacked the victim. In his appeal, he claims that: (1) the trial court improperly denied his motion for a new trial in light of the exculpatory blood typing and DNA profiling evidence he produced at trial; (2) the state’s attorney made improper arguments to the jury, *267resulting in a violation of his right to a fair trial before an impartial jury; and (3) the trial court improperly denied his posttrial motion for blood testing and DNA profiling of vaginal swabs and smears from the victim.
I
The first issue in this appeal is whether the verdict of the jury was so clearly against the weight of the evidence that the trial court improperly denied the defendant’s motion to set aside the verdict and for a new trial. After reviewing the standard by which we review the trial court’s disposition of such a motion, we will consider in turn the inculpatory evidence offered by the state and the exculpatory evidence offered by the defendant.
A
The defendant’s motion for a new trial differs from a motion for acquittal in that it does not dispute that the state presented sufficient evidence, if found credible by the jury, to sustain his conviction. The defendant maintains, instead, that the exculpatory evidence he offered in his defense was so strong that the trial court improperly refused to rule that the verdict was contrary to the manifest weight of the evidence.
Although the jury is ordinarily the sole arbiter of the facts in a criminal case, its power “is not absolute. . . . The court serves a supervisory function vis-a-vis the jury. ... ‘In passing upon a motion to set aside a verdict, the trial judge must do just what every juror ought to do in arriving at a verdict. The juror must use all his experience, his knowledge of human nature, his knowledge of human events, past and present, his knowledge of the motives which influence and control human action, and test the evidence in the case according to such knowledge and render his verdict accordingly. A juror who did not do this would be remiss in *268his duty. The trial judge in considering the verdict must do the same, or fail in the discharge of that function which the law has laid upon him; and if, in the exercise of all his knowledge from this source, he finds the verdict to be so clearly against the weight of the evidence in the case as to indicate that the jury did not correctly apply the law to the facts in evidence in the case, or were governed by ignorance, prejudice, corruption or partiality, then it is his duty to set aside the verdict.’ Howe v. Raymond, 74 Conn. 68, 72, 49 A. 854 (1901).” State v. Avcollie, 178 Conn. 450, 456-57, 423 A.2d 118 (1979), cert. denied, 444 U.S. 1015, 100 S. Ct. 667, 62 L. Ed. 2d 645 (1980), aff’d, 188 Conn. 626, 453 A.2d 418 (1982), cert. denied, 461 U.S. 928, 103 S. Ct. 2088, 77 L. Ed. 2d 299 (1983). In such a case, “[a] verdict may be set aside even if the evidence was conflicting and there was direct evidence in favor of the party who prevailed with the jury.” Palomba v. Gray, 208 Conn. 21, 24, 543 A.2d 1331 (1988).
One cogent reason for overturning the verdict of a jury is that the verdict is based on conclusions that are physically impossible. “[A] verdict should be set aside ‘[wjhere testimony is thus in conflict with indisputable physical facts, the facts demonstrate that the testimony is either intentionally or unintentionally untrue, and leave no real question of conflict of evidence for the jury concerning which reasonable minds could reasonably differ.’ Budaj v. Connecticut Co., 108 Conn. 474, 476, 143 A. 527 (1928).” State v. Avcollie, supra, 178 Conn. 457. We have found the requisite physical impossibility in cases involving the presence of adequate lighting, which contradicted an accident victim’s testimony that he did not and could not see a trolley bearing down on him. Budaj v. Connecticut Co., supra, 475-76. In Gianotta v. New York, N.H. & H. R. Co., 98 Conn. 743, 744, 120 A. 560 (1923), we held that a tort claimant could not reasonably have failed to see *269an approaching train in light of his own testimony that he had stopped his car some fifteen feet from the track.
Scientific evidence is relevant to a determination of what is physically impossible. In Roma v. Thames River Specialties Co., 90 Conn. 18, 20-21, 96 A. 169 (1915), this court held that the trial judge “would have failed in his duty” if he had not set aside the verdict when “the laws of mechanics, as testified to and uncontradicted, tended to prove [the claimant’s] story impossible.” In Jump v. Ensign-Bickford Co., 117 Conn. 110, 116, 167 A. 90 (1933), the trial court properly set aside the verdict when expert scientific testimony indicated that it was physically impossible for a fuse to burn as fast as the claimant had alleged, and this court could “find in the evidence no reasonable ground which would have justified the jury in disregarding that evidence.” In Wadlund v. Hartford, 139 Conn. 169, 171-72, 91 A.2d 10 (1952), this court held that the trial judge should have ordered a new trial where the indisputable physical facts established by the records of the weather bureau conclusively determined the incredibility of testimony that a sidewalk had been icy for at least three days before the claimant’s slip and fall injury. As a general matter, we have held that “[wjhere the trial court rejects the testimony of [an] expert, there must be some basis in the record to support the conclusion that the evidence of the [expert witness] is unworthy of belief.” (Internal quotation marks omitted.) Builders Service Corporation v. Planning & Zoning Commission, 208 Conn. 267, 294, 545 A.2d 530 (1988).
Appellate review of a trial court’s decision granting or denying a motion for a new trial must take into account the trial judge’s superior opportunity to assess the proceedings over which he or she has personally presided. Although our decisions have not been entirely consistent in delineating the appropriate standard for *270appellate review, we have most often characterized the action of the trial court as an exercise of its legal discretion and have accordingly confined our role to a determination of whether there has been an abuse of discretion. Palomba v. Gray, supra, 24-25; Lee v. Lee, 171 Conn. 1, 2-3, 368 A.2d 11 (1976);Roma v. Thames River Specialties Co., supra, 20. Although we have occasionally taken a broader view; State v. Avcollie, supra, 178 Conn. 460; Burr v. Harty, 75 Conn. 127, 129-30, 52 A. 724 (1902); we now conclude that the appropriate standard of review is abuse of discretion.2
B
Our determination of whether the trial court abused its discretion in this case must begin with a review of the state’s evidence against the defendant. The state presented substantial evidence in support of the conviction from the victim, from other witnesses who corroborated her observations and from the conduct of the defendant himself.
The defendant does not dispute that, on November 30,1987, the victim was struck, kidnapped and then sexually assaulted by a man who first accosted the vic*271tim as she was walking along a dark stretch of road in the vicinity of her home. The initial attack on the victim knocked off her eyeglasses, without which her vision is approximately 20/375 and she has difficulty seeing anything more than three feet away. The incident took place at night, and the victim testified that “[i]t was very dark; it was pitch black.” Nonetheless, the victim had seen the man when she first turned around as he was running up behind her, and she had other opportunities to see him close up when he knocked her down, when he forced her into his car, when he drove her to the site of the attack, and when he sexually assaulted her. She testified that her view of his face was then “clear” and “plain.”
Although she was unable to see where her assailant was driving her, the victim thought they had driven for ten to fifteen minutes, eventually turning onto what she deduced to be a dirt road. The road seemed bumpy to her, and she could hear tree branches or brush scraping the sides of the car. Because the man drove directly to the spot and turned onto the dirt road without hesitation, she concluded that he was familiar with the area.
The victim was able to give a detailed description of her assailant’s car, which she recognized as a greenish blue Chevrolet sedan. She noticed a watch hanging on the car’s automatic gear shift. She also saw that the car had an infant car seat in the back with what appeared to be white patches on it, that the car was in general disarray, and that it was filled with tools and “junk.” As she left the car, she noticed that the door handles were silver levers that had to be lifted up to open.
In her first meeting with police, she described her assailant as a twenty year old white male, between 5' 6" and 5' 8", 140 pounds, with short, brown, messy hair and a light moustache. With the help of Detective Roy *272Bredefeld, and using overlay feature foils, the victim made a composite sketch of her assailant. She characterized the composite as “very, very similar” to the perpetrator’s actual appearance. The police released the composite sketch to local newspapers. The sketch resembled the defendant.
About two weeks after the incident, the police received information indicating that the defendant was responsible for the assault. When Bredefeld went to question the defendant at his home, which was only a quarter of a mile from the place where the assailant had released the victim, Bredefeld noticed that the defendant had what appeared to be a week and one half to two week old beard.
The defendant denied having attacked the victim. He initially told Bredefeld that he was home alone watching television on the night of the assault. He later changed his alibi several times, claiming variously that he was out with his friend Andy Neill or working late. He also stated that he had genital warts and therefore could not have sex with anyone. When the police asked him for the name of a doctor who could corroborate this claim, he failed to provide one. The defendant admitted that he owned a 1976 Chevrolet Malibu, which fit the victim’s description, was registered in his wife’s name, and was parked in a spot assigned to him, but he claimed that he had not driven the car in about a month.
In mid-December, Bredefeld assembled two photo arrays. The first contained a black and white photograph of the defendant taken several years earlier. The second contained a more recent color photograph of the defendant with a beard. In compiling both arrays, the officer made a conscious effort to include photos of young men resembling the defendant’s general appearance.
*273Detective Mark Schultz went to the victim’s home and showed her the first array. Although the victim immediately and positively selected the defendant’s photograph as that of her assailant, she said that he looked younger in the photograph and asked to see a more recent photograph of him. When Schultz then showed her the second array, she once again quickly selected the defendant’s photograph, but noted that he had not had a beard at the time he had assaulted her.
On March 5,1988, the victim again went to the police station, where the police showed her an array that included a more recent photograph of the defendant portraying him without a beard. Again choosing the defendant’s photograph, she said that she had no doubt that he was the man who had assaulted her. Finally, she reaffirmed her positive identification of the defendant by identifying him in the courtroom.
On December 17,1987, the police executed a search and seizure warrant for the defendant’s car. When the police attempted to execute the warrant, the defendant attempted to prevent the officers from taking the car by blocking it in with another one of his vehicles. The police were nevertheless able to search the car. The appearance of the car, a greenish blue 1976 Chevrolet Malibu, fit the victim’s description of the perpetrator’s car in several respects. It was indeed “messy,” as the victim had described it. The door handles and gear shift conformed to her description. In the back seat was an infant seat with rips in the upholstery, exposing white stuffing.3 The car’s right fender area *274had several horizontal scratches, the longest of which was approximately seven inches. There was also a significant amount of mud and dirt on the lower rocker panel or wheel well of the vehicle.
Finally, the police found several human head hairs on the left and right front seats, left front floor, and right rear floor of the defendant’s car. Later forensic examination revealed that these hairs resembled the victim’s hair and could have come from the victim. The expert witness who had examined the hairs testified, however, that hair matching is so far from being a positive means of identification that it is “the weakest link” in the police laboratory’s array of tests, that the hair in the car might have come from any other person with similar hair, and that the hair might have been in the car for five years before the police found it.
When the police confronted the defendant with the fact that several hair samples taken from his car were similar to the victim’s head hairs, he told them that he might have picked her up hitchhiking. The victim later identified the car, in photographs, as the one into which she had been forced.
Two of the defendant’s former friends testified for the state. According to Valerie Hilton, who had been the defendant’s girlfriend at the time of the assault, the defendant brought a newspaper article about the incident to her attention on December 1,1987. He told her that although he fit the description, he had been with his friend Andy Neill that night and was not responsible. He brought up his similarity to the description several times. She was so frightened by his “obsession” with the incident that she terminated their relationship.
Hilton further testified that sometime after he had brought the article to her attention, she and the defendant were leaving a downtown building when they *275noticed a group of police officers outside. The defendant became very nervous and told Hilton that he was afraid the police were going to arrest him because of the incident in the paper. While still maintaining his innocence, he urged her to run into nearby woods with him to avoid the officers. Hilton remembered seeing the defendant driving his 1976 Chevrolet Malibu around the time of the assault, contrary to the defendant’s statement that he had not driven that car for a month.
Andy Neill, who had been the defendant’s friend at the time of the incident, testified that the defendant had asked him once or twice whether he remembered that the two of them had been together on the night of the assault. Neill had told him that he did not recall whether or not they had been together. Neill also testified that the defendant had a habit of hanging his wrist watch on the gear shift of the car when he drove.4 On cross-examination, Neill admitted that he was angry at the defendant and had broken off their friendship.
In its totality, this evidence presented the jury with a strong case for the culpability of the defendant. Absent strong exculpatory evidence, a reasonable juror could have found the defendant’s guilt established beyond a reasonable doubt by the victim’s consistent positive identification of the defendant, her substantially accurate description of his car and its contents, the scratches on his car consistent with its being driven into the woods as described by the victim, the numerous actions by the defendant that appeared to display a consciousness of guilt, and the testimony of one former friend of the defendant that the defendant had a habit of hanging his watch on the gear shift of his car, as the assailant had done.
*276c
Despite the strength of the state’s case, the defendant argues that he presented scientific exculpatory evidence that was so strong that the trial court abused its discretion in concluding that the jury’s verdict was not contrary to the manifest weight of the evidence.5 This evidence consisted of chemical analyses of semen stains found on the victim’s clothing. These scientific tests, consisting of blood grouping results and DNA profiles, indicated that the defendant could not have been the source of the semen stains.
1
The victim was examined at Bristol Hospital promptly after the assault. There she removed her blue jeans and underpants, which were sealed at once in an evidence bag. Her underpants had been clean when she put them on, on the morning of her assault; she had changed into her blue jeans when she had gotten home from work approximately one hour before the assault. Gary Miller, the examining physician, examined her, took combings of her pubic hair, and used cotton swabs to take samples from her mouth and vagina. One vaginal swab was immediately tested for semen, and tested positive. Miller concluded that sexual activity had taken place. The victim testified that she had not had sexual relations with anyone other than her assailant that day.6
*277At the hospital, a second vaginal swab was used to put vaginal smears on two glass slides. The second vaginal swab and two oral swabs were then sealed within the plastic box of a standard “rape kit” along with the slides, the hair combings, and an orange stick with blood that had been taken from the victim’s ear. The sealing process was witnessed by a nurse. Both Miller and the nurse signed the tape on each sealed package. Miller, who had examined sexual assault victims many times before, testified that he consciously secured the samples for the rape kit and the clothing so that there would be no break in the chain of custody. The next day, Detective Rodney Gotowala delivered the sealed rape kit and evidence bags intact to the state toxicological lab in Hartford, where they were logged in and placed in a secure vault.
On December 8,1987, Sanders Hawkins, chief state toxicologist, signed out these materials. He broke the seal on the rape kit, removed the swabs and slides, clipped off a piece of the vaginal swab to test it for acid phosphatase, an enzyme unique to semen, and obtained a positive result. He also microscopically examined one of the slides containing a vaginal smear, and found sperm on it. The swabs and slides were then put back in the rape kit, which was resealed, returned to the evidence vault, and stored until trial.
Hawkins also examined the clothing in the evidence bags. Illuminating each piece of clothing with an ultraviolet light, he found a stain on the inside crotch of the underpants and another on the lower left leg of the blue jeans. After the stains had been outlined with a red crayon, they tested positive for acid phosphatase, and thus for semen. The stain on the blue jeans was found to contain A and H antigens, indicating that a secretor with type A blood had contributed to it. Nei*278ther the underwear, the vaginal swab, nor the vaginal smear were tested for blood groupings at that time, because the laboratory’s policy was not to blood type stains of mixed origin for fear of confusing investigators or giving false leads.7
On January 6, 1989, again at the state’s request, Hawkins tested the underwear stain and retested the blue jeans stain for blood groupings. Both stains tested positive for A and H antigens, consistently indicating that at least one contributor to each stain was a secretor with type A blood. Elaine Pagliaro of the state police forensic laboratory, who received both the underwear and the blue jeans in March, 1989, obtained identical results.
Pagliaro also tested whole blood and saliva samples from the defendant, the victim, and the victim’s boyfriend. She determined that each was a secretor with type 0 blood. Since persons with type 0 blood lack A or B antigens, and secretors with type 0 blood secrete only H antigens, she concluded that someone other than they must have contributed to the stains.
Lawrence Presley, a forensic analyst from the Federal Bureau of Investigation’s DNA analysis unit in Washington, D.C., tested the samples in May, 1989, at the state’s request. Presley used standard DNA profiling techniques to compare certain unique genetic characteristics in DNA extracted from the semen stains on the victim’s underwear and blue jeans with those of DNA extracted from known blood samples from the victim, her boyfriend, and the defendant. Although the semen stains on the blue jeans had degraded to the point at which no reliable DNA comparisons could be made, Presley concluded that neither the defendant nor the victim’s boyfriend could have contributed any part of the semen stain on the victim’s underwear.
*2792
It is physically impossible for the defendant to have been the man who sexually assaulted the victim unless both the blood typing tests and the DNA tests made on the samples taken from the victim’s clothing are for some reason unreliable. Unreliability may arise from inherent flaws in the testing procedures themselves, or from contamination of the sample. The state has suggested both of these possibilities as an explanation for the test results in this case.
Blood typing tests, at least insofar as they exclude someone from sexual contact, are generally acknowledged to be absolutely reliable when the sample is not contaminated. In paternity proceedings, where they are most often used, their reliability is so well established as to be properly subject to judicial notice and their results are given conclusive weight in most states. See Little v. Streater, 452 U.S. 1, 12, 14, 101 S. Ct. 2202, 68 L. Ed. 2d 627 (1981); Moore v. McNamara, 201 Conn. 16, 26-27, 513 A.2d 660 (1986); 1 S. Schatkin, Disputed Paternity Proceedings (4th Ed. Rev. 1991) §§ 9.04, 9.13 ; Uniform Laws Annotated, Uniform Act on Paternity (1987) § 10. General Statutes § 46b-168 (a) makes a blood test admissible only if the court finds, as a matter of law, that the test results “establish definite exclusion of the putative father . . . .” Once a court has made a finding allowing such exculpatory evidence to be admitted, a jury would not be permitted, in disregard of that evidence, to find a man to be a child’s father.
Because exclusionary blood typing evidence is dis-positive against an allegation of paternity, which need only be proven by a preponderance of the evidence; see Palomba v. Gray, supra, 25; it would be anomalous to give it less weight here, where the state must prove *280the defendant’s guilt beyond a reasonable doubt. The only possible distinction between the two cases is a difference in the risk of contamination. In paternity cases, the tested samples are taken in sterile laboratory conditions. That is not the case here or in criminal cases generally. Without having offered any expert evidence to counter the defendant’s scientific evidence, the state urges that the risk of contamination in this case undermines the weight of the scientific inferences on which the defendant relies.
Given the undisputed testimony that the victim’s underwear had been put on clean the morning of the incident, that she had changed into the blue jeans shortly before the incident, and that the samples had been handled with great care afterward, any contamination seems improbable. Such contamination would have had to take place in the laboratory or when the samples were in police custody. Moreover, both contaminants would have to have been bodily fluids from a secretor with type A blood, and the contaminants would have to have been carefully applied only within the crayon outline of the original stain on the clothing that was made by Hawkins when he first examined the clothing.
The results of the DNA testing, moreover, confirmed the results of the blood type testing. DNA profiling is concededly a procedure so novel that its reliability is still debated. Most of the debated issues are not, however, relevant to this case.8
*281The basic theory of DNA testing, which Presley explained at trial, is the following. DNA, the fundamental genetic material of all animals and plants, is composed of two parallel chain-like structures, each segment of which is a pair of chemical bases. The sequence of the base pairs determines each individual’s genetic traits. Most DNA does not vary from person to person, but certain parts of the DNA, called “polymorphic loci,” vary distinctly from one person to another.
“Restriction enzymes” are able to recognize a specific sequence of base pairs along a single strand of DNA, and to cut the DNA at that location. This enables the researcher to conduct restriction fragment length polymorphism (RFLP) analysis, which involves a number of steps: (1) extraction of the DNA from the sample; (2) fragmentation of the DNA by restriction enzymes; (3) gel electrophoresis, in which the fragmented sample is placed upon an electrically charged *282agarose gel, where it forms a distinctive pattern while moving from the negative to the positive charge; (4) Southern blotting, in which the DNA pattern is transferred to a more durable nylon membrane; (5) hybridization, in which DNA patterns specific to the individual are revealed by the use of radioactive probes designed to seek out and bond with polymorphic regions of the DNA; and (6) autoradiography, in which a piece of film is developed on top of the nylon membrane, revealing the location of the radioactive probes by bands on the film. Because the pattern of male DNA found in the sample did not match the defendant’s, Presley concluded that the defendant could not be the man who had attacked the victim.
Some commentators, however, have questioned the reliability of DNA testing even with respect to a clean sample. “The art to the technique is in knowing that: (1) each sample contains the same amount of DNA; (2) the gel is the exact percentage required; (3) the buffer is the proper ionic strength; (4) the gel is not poured too cool; (5) the wells [of gel] are formed properly because the comb [determining their shape] was not removed too soon; and, (6) the voltage is correct.” L. Adkison, “DNA Fingerprinting: A Scientific Perspective,” 42 Mercer L. Rev. 1099, 1102 (1991). If the proper protocols are not followed, the test can yield misleading results, producing incorrect bands on the DNA print. See W. Thompson & S. Ford, “DNA Typing: Acceptance and Weight of the New Genetic Identification Tests,” 75 Va. L. Rev. 45, 64-70 (1989); see also E. Imwinkelried, “The Debate in the DNA Cases Over the Foundation for the Admission of Scientific Evidence: The Importance of Human Error as a Cause of Forensic Misanalysis,” 69 Wash. U. L. Q. 19 (1991).
In this case, however, it does not appear that the distortion that has concerned the commentators could have occurred, since the female portion of the DNA that *283came from the stain matched “exactly” with that of the victim. Presley testified that such a match indicated that the test had been conducted properly.
The state has made two arguments supporting its theory that the samples were contaminated. First, before the jury the state maintained that there was testimony from which it could be inferred that something had been added to the samples before May, 1989, when the DNA tests were conducted. Second, in this court the state argues, for the first time, that detergent traces or other contaminants might have altered the DNA in the sample so that it was no longer recognizable as that of the defendant.
The state’s argument to the jury that such contamination had occurred is difficult to square with the unchallenged testimony that the clothing was, after the victim’s attack, meticulously preserved in accordance with the requirements of maintaining a clear chain of custody. The state relies, however, on the testimony of two police detectives, who allegedly were told by Hawkins in January, 1988, that there was insufficient stain left on the samples to perform further tests. At trial, Hawkins could not recall having made that statement. When evidence was offered before the jury that further testing was in fact performed three times in January, 1989, the state argued that this inconsistency indicated that body fluid had been added to the stain in the interim.9
The state’s theory that contamination occurred sometime between January, 1988, and January, 1989, does *284not, however, explain the source of the A antigens that were found in the blue jeans stain on December 8,1987, a month before the alleged conversation between Hawkins and the police. In order for the defendant to be the source of the semen, there would have had to have been another, earlier episode of contamination in which the A antigens were added. Given the uncontradicted evidence that the sample was in secure, clean storage throughout this period, even a single episode of contamination seems most unlikely. The state’s theory requires two such episodes, during at least one of which both samples were contaminated.
An alternative, more parsimonious theory (not argued by the state) is that there was only a single episode of contamination, which occurred before December 8, 1987. There is, however, no evidence of contamination at that time. Moreover, the conversation cited by the state in its closing argument, which is the only evidence of contamination upon which the jury could have relied, would then be conceded to be of no probative value whatsoever. “Inferences to be drawn from the facts must be reasonable and logical, and must not be the result of guess, conjecture or speculation.” Toomey v. Danaher, 161 Conn. 204, 211, 286 A.2d 293 (1971); see also Boehm v. Kish, 201 Conn. 385, 389, 517 A.2d 624 (1986).
The state’s theory also fails to explain how the DNA test could have falsely exculpated the defendant. Even if someone else’s blood, semen, or saliva had been added to the small patch of clothing that had been circled with a red crayon, the state has not suggested how such bodily fluids could have destroyed the DNA that was already present in the stain.10 That could only have been *285done by some other chemical which acts as a denaturing agent. Any contamination theory thus requires two distinct contaminants, one containing A antigen and the other containing the denaturing agent.
The state’s second theory, offered to this court but not to the jury, is that a denaturing agent might have been present in the clothing before the victim put it on and might have changed the sample to produce falsely exculpatory results. According to the scientific literature, the state may be correct that it is physically possible for residues of detergent in clothing to produce that result, and the defendant may be incorrect in the claim made during his rebuttal at oral argument that such contamination could only produce an unreadable smear. See W. Thompson & S. Ford, supra, 66, 68, 87 n.188; J. Hoeffel, “The Dark Side of DNA Profiling: Unreliable Scientific Evidence Meets the Criminal Defendant,” 42 Stan. L. Rev. 465, 481-82 (1990); but see L. McNally et al., “Evaluation of Deoxyribonucleic Acid (DNA) Isolated from Human Bloodstains Exposed to Ultraviolet Light, Heat, Humidity, *286and Soil Contamination,” 34 J. Forensic Sci. 1059 (1989) (concluding that contaminants did not alter the integrity of DNA and no false RFLP patterns were obtained); L. McNally et al., “The Effects of Environmental and Substrata on Deoxyribonucleic Acid (DNA): The Use of Casework Samples from New York City,” 34 J. Forensic Sci. 1070 (1989) (reliable tests performed on evidentiary items with unknown environmental history and dried onto a variety of substrata).
On the present record, however, the state’s theory of postassault contamination is untenable. The same stain on the victim’s underpants that produced a reading of DNA that was identifiably not that of the defendant also produced a reading of DNA that matched the victim’s “exactly.” The state has not explained how it could be physically possible for a chemical contaminant selectively to alter the defendant’s DNA to the point at which it was no longer recognizably his, while leaving the victim’s DNA perfectly intact. The only testimony of record that addresses the question of contamination is that of Presley, who, upon cross-examination by the state, stated that he “did not see any severe degradation” of the sample.
3
Had all these matters been plainly put to the trial court, the court might well have concluded that rebuttal of the defendant’s scientific evidence on a theory of contamination must rest not only on speculation, but on a scenario in which physically impossible events occur. If indeed “the state’s theory was based on physically impossible facts and conclusions”; State v. Avcollie, supra, 178 Conn. 469; then the defendant’s motion for a new trial should have been granted. This exculpatory evidence raises a serious question “ ‘whether the jury could reasonably have concluded, upon the facts established and the inferences reasonably drawn *287therefrom, that the cumulative effect of the evidence established guilt beyond a reasonable doubt.’ State v. Chetcuti, 173 Conn. 165, 172, 377 A.2d 263 (1977).” State v. Avcollie, supra, 178 Conn. 457-58.
This was surely a hard case for the trier of fact. While, as we have seen, certain improbable speculations are necessary to support a theory of guilt, equally improbable speculations are necessary to support a theory of innocence. A trier of fact could discount the victim’s identification of the defendant as mistaken in light of her poor vision and the darkness. The defendant’s strange behavior, which seems so probative of consciousness of guilt, might be explained as the panicked and not very intelligent response of an innocent man after seeing in the newspapers a police sketch of the assailant resembling him. Because Neill admitted that he was angry at the defendant, the jury might have disbelieved his testimony that the defendant was in the habit of hanging his watch on the gear shift of his car, as the assailant had done; another witness who had ridden in the defendant’s car many times had never noticed such a habit. Doubtless the defendant’s car is not the only dark colored Chevrolet in Connecticut that has a “messy” interior or scratches, mud, and dirt on its exterior. Nonetheless, it would be strange for the defendant coincidentally to resemble the assailant so closely that the victim could correctly observe that he was younger in the police photograph than he had been at the time of the attack; cf. State v. Tillman, 220 Conn. 487, 502, 600 A.2d 738 (1991); or for him to happen to have a child safety seat in his back seat with rips in it just like the ones noticed by the victim.
When the record reveals that the theory of guilt and that of innocence both require equally improbable speculations, our constitutional jurisprudence requires weight to be given to the presumption of the innocence of the defendant. The trial court properly instructed *288the jury that “any conclusion reasonably drawn from the evidence, which is consistent with the innocence of the accused, must prevail. And the accused given the benefit of such conclusion. Proof beyond a reasonable doubt is proof which precludes every reasonable hypothesis except guilt, is consistent with guilt, and is inconsistent with any other reasonable conclusion. If you can in reason reconcile the facts proven with any reasonable theory consistent with the innocence of the accused, then you cannot find him guilty.” See State v. Little, 194 Conn. 665, 671-72, 485 A.2d 913 (1984). It follows that if the hypothesis consistent with guilt and that consistent with innocence are equally reasonable, because they are equally unreasonable, then the defendant’s conviction cannot stand.11
On these facts, the trial court would clearly not have abused its discretion if it had granted the motion for a new trial. Whether it abused its discretion in failing to grant the motion is a more difficult question. It is not, however, necessary for us to decide this question today.
The record indicates that the trial court was not adequately apprised of the contours of the issue with respect to which it was asked to exercise its discretion. The logical inconsistencies that we have found in the prosecution’s theory of the case were not brought to the court’s attention by counsel for the defense. It is possible that, fully informed, the trial court might have ruled differently on the motion for a new trial. Because *289its ruling implicates the constitutional right to a presumption of innocence and the defendant’s right to a fair trial; State v. Golding, 213 Conn. 233, 567 A.2d 823 (1989); State v. Evans, 165 Conn. 61, 327 A.2d 576 (1973); we therefore remand this case to the trial court for reconsideration of the defendant’s motion to set aside the verdict in light of the considerations discussed herein.
II
The defendant’s second claim is that the conduct of the state’s attorney, in his summation to the jury, violated the defendant’s right to a fair trial. The defendant maintains that the state’s attorney: (1) argued facts not in evidence; (2) invited speculation; (3) asserted his own opinion; and (4) promoted a verdict based on passion and emotion. None of these claims was preserved at trial. In order to be reviewable by this court, they must therefore meet the standards set forth in State v. Golding, supra, and State v. Evans, supra, which require, inter alia, that the defendant demonstrate that he has clearly been deprived of a fundamental constitutional right and a fair trial. Most of these claims do not implicate fundamental constitutional rights.
The most serious of the defendant’s claims of prosecutorial misconduct is that the state’s attorney, on several occasions during his summation, stated his personal belief in the credibility of the victim’s story and in the contamination of the test samples on which the defendant’s scientific analysis relied. As the state now acknowledges, such statements are clearly improper. State v. Williams, 204 Conn. 523, 541-44, 529 A.2d 653 (1987). In context, however, they were not “so prejudicial that [the defendant] ‘was deprived of the opportunity for a fair trial and [that] the entire proceedings were tainted.’ State v. Hawthorne, 176 Conn. 367, 372, 407 A.2d 1001 (1978).” State v. Richardson, 214 Conn. 752, 760, 574 A.2d 182 (1990).
*290“ ‘[I]n the cases in which we have denied Evans review, the claimed misconduct consisted of isolated and brief episodes, and did not reveal a pattern of conduct repeated throughout the trial. Moreover, the misconduct was not blatantly egregious.’ [State v. Williams, supra,] 537.” State v. Tweedy, 219 Conn. 489, 509, 594 A.2d 906 (1991). “Our careful review of the record in this case reveals that the prosecutor’s challenged conduct, while improper in certain instances, was ‘unrepresentative of a pattern of conduct repeated throughout the trial.’ State v. Castonguay, 218 Conn. 486, 507, 590 A.2d 901 (1991).” Id. The defendant cannot succeed on this claim of impropriety.
Ill
Finally, the defendant claims that the trial court improperly denied his posttrial motion for DNA profiling and blood testing of vaginal swabs and smears that were taken from the victim on the day she was attacked. The trial court denied this motion because the testing would produce evidence that was merely cumulative and because the motion was untimely.
Although, prior to the trial, the fluid on the vaginal swab and smears in the rape kit might have seemed a relatively unimportant piece of evidence, since other samples taken from the victim’s clothing had been extensively tested, the importance of those samples grew during the trial. The state bolstered its theory of contamination by repeatedly questioning the evidentiary integrity of the tested stains in light of the available, testable, carefully preserved, but still untested semen on the swab and smears in the kit. The state elicited testimony from the expert witnesses that the tested clothing stains were located on materials outside the victim’s body onto which contaminating substances from third persons might at least in theory have been dripped, dropped, thrown or otherwise trans*291ferred prior to testing. Had such contamination occurred, Hawkins conceded, the results at least of the blood grouping would be inconclusive, since the antigens in the sample could not be attributed to a single individual. The samples in the rape kit, by contrast, contained fluids that came directly from inside the victim’s vagina, where no comparable risk of contamination existed. Each expert acknowledged that any sample taken directly from the victim is always preferable for testing purposes to a sample extracted from a stain.
Defense counsel claimed at the sentencing hearing, and indicated that he would be prepared to testify under oath, that during the trial he had spoken with Hawkins, an employee of the state, who had told him that the entire sample in the rape kit had been used up so that no blood groupings or DNA tests could be done on it. After trial, counsel learned from three different persons that such testing might be possible. He therefore moved for posttrial discovery requesting that the testing be done by the state police forensic laboratory.
The state opposed the motion arguing that there is no legal basis for a motion for posttrial discovery. The state also argued that the defendant should be held to what the state claimed had been his tactical decision not to have the sample tested before trial. The trial court denied the motion reasoning that because the scientific evidence had been largely exculpatory anyway, the evidence he sought would be cumulative. The court also suggested that defense counsel was at fault in making his motion at that time because “this could have been done in advance of trial.” The defendant excepted and on appeal claims that the court improperly denied his motion.
The state argues, as it did at the trial, that the defendant made a tactical choice not to have the samples tested prior to the verdict. This argument presupposes
*292that it reasonably appeared during the trial that the samples were testable, and raises the inference that the state also made a tactical choice not to have the samples tested prior to the verdict.12 Such a tactical choice would plainly have been a breach of the prosecutor’s ethical duty to pursue relevant evidence even if it may be exculpatory. “It is unprofessional conduct for a prosecutor intentionally to avoid pursuit of evidence because he or she believes it will damage the prosecution’s case or aid the accused.” 1 A.B.A. Standards Relating to the Prosecution Function § 3-3.11 (c). “It is well established . . . that a state’s attorney has a duty, not solely to obtain convictions, but ‘to ensure that all evidence tending to aid in the ascertaining of the truth be laid before the court, whether it be consistent with the contention of the prosecution that the accused is guilty.’ State v. Moynahan, 164 Conn. 560, 568, 325 A.2d 199 [1973] . . . .” State v. Mitchell, 169 Conn. 161, 165-66, 362 A.2d 808 (1975).
The state seeks to avoid this difficulty by also claiming, inconsistently, that “if prior to trial Dr. Hawkins did not believe there was enough of a sample on the swabs or smears to test, and even after the verdict would admit to only a possibility that they could not be tested, the state should not be considered derelict in not ordering a test whose feasibility was either unknown or in grave doubt.” This, however, impales *293the state on the other horn of a dilemma. Either both sides were entitled to believe Hawkins’ statement or neither was. If both attorneys were entitled to believe the statement, and in fact the statement was false, then the state has breached its duty to disclose exculpatory evidence. See Brady v. Maryland, 373 U.S. 83, 87, 83 S. Ct. 1194, 10 L. Ed. 2d 215 (1963); State v. Cohane, 193 Conn. 474, 495-96 n.16, 479 A.2d 763, cert. denied, 469 U.S. 990, 105 S. Ct. 397, 83 L. Ed. 2d 331 (1984); State v. Thomas, 245 N. J. Super. 428, 434, 586 A.2d 250 (1991) (dictum that state may have an obligation to test evidence for DNA patterns); see also General Statutes § 54-86c; Practice Book § 741. If Hawkins, the director of the state toxicological lab, falsely (although in good faith) denied the availability of the requested exculpatory evidence, his error must be attributed to the state. See Giglio v. United States, 405 U.S. 150, 92 S. Ct. 763, 31 L. Ed. 2d 104 (1972); Fulford v. Maggio, 692 F.2d 354, 358 n.2 (5th Cir. 1982); Barbee v. Warden, 331 F.2d 842, 846 (4th Cir. 1964); State v. Gunning, 183 Conn. 299, 307, 439 A.2d 339 (1981); see also United States v. Bagley, 473 U.S. 667, 105 S. Ct. 3375, 87 L. Ed. 2d 481 (1985).
In this case, the trial court held that the evidence sought by the defendant would be cumulative. Ordinarily, that would dispose of this issue, and it might also dispose of the alternative claim that the prosecutor had breached his duty to pursue all relevant evidence. In assessing the legal consequences of possible Brady violations, a court appropriately considers the probable effects of the nondisclosure on the result of the trial. The state’s failure to disclose exculpatory information will require a new trial “ ‘only if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different. A “reasonable probability” is a probability sufficient to undermine confidence in the outcome.’ ” *294State v. Shannon, 212 Conn. 387, 399, 563 A.2d 646, cert. denied, 493 U.S. 980, 110 S. Ct. 510, 107 L. Ed. 2d 512 (1989), quoting United States v. Bagley, supra, 682. “ ‘There is a difficulty inherent in measuring the effect of nondisclosure in the course of a lengthy trial with many witnesses and exhibits . . . this lack of certitude suggests deference by a reviewing court especially in the weighing of evidence.’ [State v. Pollitt, 205 Conn. 132, 149, 531 A.2d 125 (1987)] .... Thus, the trial court’s decision regarding the defendant’s petition for a new trial as a result of a Brady violation will be overturned only upon a finding of ‘clear abuse of discretion.’ Demers v. State, [209 Conn. 143, 148, 547 A.2d 28 (1988)] . . . .” State v. Shannon, supra, 400.
In this case, however, there is reason to doubt that the evidence would be cumulative. The state, and perhaps also the jury, laid great weight on the fact that samples taken from clothing are, as a general matter, more susceptible to contamination than those taken directly from the body of the victim. The state also claimed at trial that there was tangible evidence that contamination of the clothing samples had occurred, but as we have seen in part I of this opinion, supra, that claim was unfounded. The results of a new test are therefore important either to support or further to discredit the jury’s verdict of guilty.
The state correctly observes that “[t]he vaginal swabs and smears, while perhaps less susceptible to contamination in the gathering process, are just as susceptible as any other evidence to contamination in handling and storage.” Nonetheless, if this evidence were to exculpate the defendant, it would become necessary for the state to claim, on the basis of no evidence, a minimum of three separate episodes of contamination of three different samples in the secure conditions of the police laboratory. Under such circumstances, the *295defendant’s case for a new trial would become stronger. It is not clear that a jury would, or reasonably could, convict a defendant on the basis of such unfounded and improbable speculation.
The trial court was not fully apprised of these considerations when it ruled that the evidence would be cumulative. We will not consider whether the trial court abused its discretion with respect to the discovery motion for the same reasons that we have not reached the question whether the trial court abused its discretion in denying the motion for a new trial. When it made both these decisions, it was not aware of the logical inconsistencies in the prosecution’s case, the evidence suggesting that the chemical alteration of the assailant’s DNA sample was physically impossible, or the absence of any evidence that the defendant’s scientific tests were unreliable. The trial court should therefore have the opportunity to decide both of these motions in the first instance on the basis of the same considerations upon which we will rely when and if we review those rulings.13
*296The case is remanded to the trial court for further proceedings consistent with this opinion.
In this opinion Glass, Borden and Santaniello, Js., concurred.
General Statutes § 53a-92 provides in pertinent part: “(a) A person is guilty of kidnapping in the first degree when he abducts another person and when ... (2) he restrains the person abducted with intent to (A) inflict physical injury upon him or violate or abuse him sexually . . . .”
General Statutes § 53a-70 provides in pertinent part: “(a) A person is guilty of sexual assault in the first degree when such person (1) compels *266another person to engage in sexual intercourse by the use of force against such other person or a third person, or by the threat of use of force against such other person or against a third person which reasonably causes such person to fear physical injury to such person or a third person . . . ."
One commentator has noted that the trial court’s superior ability to view the trial and assess the witnesses’ credibility can be given excessive weight on review. “The assessment of credibility is not a mystical process that must be left to the intuitions of unreviewed, hopefully clairvoyant jurors. The task depends less on the ability of jurors to stare deeply into a witness’s eyes than it does on the jurors’ ability to judge the internal coherency of the witness’s story, its consistency with known external circumstances, and the witness’s past conduct, statements and character. Judicial review on the basis of a ‘cold’ record has its virtues and can safeguard at least part of this process. The opportunity to leave the emotions of the closing argument behind, to escape the interplay of courtroom personalities, and to sort out times, places, and contradictions in a relatively leisurely fashion can yield important insights.” A. Alschuler, “The Supreme Court and the Jury: Voir Dire, Peremptory Challenges and the Review of Jury Verdicts,” 56 U. Chi. L. Rev. 153, 217-18 (1989). Our presumption of the correctness of the trial court’s ruling is not irrebuttable, however, and does not prevent these virtues of appellate review from being realized.
However, the car had a white top, which had not been mentioned in the victim’s description. On cross-examination, the victim acknowledged that she had never told anyone that her assailant’s car had a white top, but explained that she could not see the top of the car because it was dark outside and the car’s top was dirty. When asked how she could tell in the darkness that the car’s top was dirty, she said that she had noticed the dirt in photographs of the car that she had seen a few weeks after the incident.
However, Tammy Woodbury, a friend of the defendant, testified that she had driven with the defendant many times and had never noticed him do anything unusual with his watch.
The defendant has raised two other grounds upon which, he maintains, the trial court should have granted his motion for a new trial. Neither of these is persuasive.
The defendant notes that the victim, whose testimony was central to the state’s case, has difficulty seeing without her glasses, and that there are inconsistencies in her description of his car. At best, these contentions go to the weight to be given the victim’s explanation of what had transpired. The defendant also produced an alibi witness who placed him in Southington shortly before the victim’s abduction. In both instances, the jury and the trial court could reasonably have reached credibility determinations adverse to the defendant.
The dissent correctly observes that neither the victim nor Miller testified to seeing stains on her clothing. Neither of them was ever asked, how*277ever, what they had seen on the clothing. The dissent’s theory that the clothing was stained at some other time does not account for the fact that semen was present in the victim’s vagina immediately after the assault.
The vaginal samples have still not been tested, and the state’s resistance to having them tested is the basis of the defendant’s third claim in this appeal. See part III, infra.
Many of the problems are associated with DNA testing by private, profit making laboratories, which provided manifestly interested witnesses who sought to establish that their procedures were accepted by the scientific community while simultaneously claiming those procedures as trade secrets and thus preventing them from being subjected to independent validation studies. See W. Thompson & S. Ford, “DNA Typing: Acceptance and Weight of the New Genetic Identification Tests,” 75 Va. L. Rev. 45, 58-60 (1989). These problems have been eliminated by the creation of the FBI laboratory, which performed the tests in this case. Other difficulties, such *281as the inherent subjectivity of interpretation when an expert declares a “match” between samples; see id., 87-89; J. Hoeffel, “The Dark Side of DNA Profiling: Unreliable Scientific Evidence Meets the Criminal Defendant,” 42 Stan. L. Rev. 465, 486-87 (1990); and the use of population statistics to estimate the likelihood of a coincidental match; see R. Lewontin & D. Hartl, “Population Genetics in Forensic DNA Typing,” Science, 20 Dec. 1991,1745; do not arise in this case, because no match was found and the mismatch was sufficiently obvious that even a layperson could verify it. “Indeed, barring technical errors resulting in false negatives, DNA mismatches are definitive with no need to invoke statistics.” Id., 1746.
None of the statements of scientific fact about DNA testing in general in this opinion should be deemed binding on Connecticut trial courts. This court does not purport to be a scientific authority, and its conclusions with respect to scientific matters are not binding on lower courts unless, as with blood tests, those conclusions are based on judicial notice of generally recognized scientific facts. Unlike some other courts; see, e.g., State v. Woodall, 385 S.E.2d 253, 260 (W.Va. 1989); we regard DNA typing as too novel for its reliability to be judicially noticed at this time. Ordinarily, with respect to scientific questions concerning what is physically possible, this court relies primarily upon expert testimony offered at trial. See, e.g., Jump v. Ensign Bickford Co., 117 Conn. 110, 116, 167 A. 90 (1933); Roma v. Thames River Specialties Co., 90 Conn. 18, 21, 96 A. 169 (1915). That is the case here.
There is some question as to whether this is a reasonable inference from the testimony, since the detectives conceded that they did not give Hawkins a laboratory number or even refer by name to the samples about which they were inquiring. The defendant argues that even if Hawkins did state that he had no more evidence that could be tested, he might have believed that the detectives were inquiring about the orange stick, which did not contain enough sample to provide a blood grouping.
The view expressed in the dissenting opinion, that the two items of clothing were somehow contaminated before the attack, is not supported by any explanation, on the record, of how or when this might have happened. The victim testified that during the attack, she kept on one leg of *285her jeans and of her underpants. Thus the most plausible explanation for the existence of the stains, both of which contained sperm, is that the ejaculate of the attacker dripped onto both items of clothing. The unspoken premise of the dissent is that both stains, which the dissent concedes to contain bodily fluids from someone other than the defendant, were deposited by some other person or persons at some time other than the time of the assault. It is indeed physically possible for the two stains each to contain the combined bodily fluids of two or more different men. The state presented nothing at trial to indicate that this physical possibility occurred.
The record is equally unclear about the possibility, to which the dissent adverts, that it is physically possible for DNA or A antigens to remain on clothing after laundering. No such claim has ever been made by the state, either at trial or on appeal. The state did not ask the expert witness whether such a possibility existed. In the absence of any challenge at trial by the state to the common sense belief that washing clothes makes them clean, the defendant apparently saw no need to introduce evidence to support that belief. It would be exceedingly harsh to penalize the defendant here for failing to introduce expert testimony to rebut a counterintuitive theory that was never mentioned during the trial.
“The burden of persuasion in an ordinary civil action is sustained if evidence induces in the mind of the trier a reasonable belief that it is more probable than otherwise that the fact in issue is true.” (Internal quotation marks omitted.) Lopinto v. Haines, 185 Conn. 527, 533, 441 A.2d 151 (1981). In a tort case, for example, if the trier of fact finds that it is equally likely that the defendant did or did not breach a duty that caused the plaintiffs harm, then the defendant must prevail. A fortiori, the same result must occur in a criminal case, where the defendant’s guilt must be proven beyond a reasonable doubt.
The prosecutor argued as follows: “[A]s far as I could tell it was his tactical decision not to seek to have that analyzed during the pendency of the case and before trial. He had a lot of evidence on his side. There is certainly good reason when you have that much going for you that he did not, to contaminate it with some contrary results that might occur from some other piece of evidence. So it is quite reasonable for Mr. Van Kirk to take the position he certainly had enough for trial and not to overdo it. This information was available.
“It’s not newly discovered evidence because it could have been—it was available to him and could have been analyzed long before the trial if counsel had tactically chosen to do that. Neither one of us did. And the result has occurred.” (Emphasis added.)
The dissent contends that the trial court had no authority to consider the defendant’s request in the context of a posttrial motion. Practice Book § 741 provides, however, that in addition to certain materials which the defendant must request within ten days after the entry of a plea, “the defendant shall be entitled to disclosure of exculpatory materials in accordance with any applicable constitutional and statutory provisions.” Practice Book § 745 provides that “[i]n addition to any other disclosure or inspection permitted by these rules, the judicial authority may, upon motion of the defendant and a showing of good cause, order such other disclosure, inspection, testing or copying of specific information and materials as the interests of justice may require.” Practice Book § 747 provides that if the prosecuting authority fails to comply with its disclosure obligations, “the judicial authority may, on motion of the defendant or on his own motion, grant appropriate relief, which may include . . . [requiring the prosecuting authority to comply . . . or . . . [ejntering such other order as he deems proper.” In light of the prosecution’s arguable failure to honor its duty to investigate or to disclose, the trial court had the authority to order immediate discovery of the rape kit in the interests of justice.