425 Mass. 819

March 5, 1997 - August 25, 1997

Bristol County


This court, examining de novo a judge's ruling, in granting a new trial in a murder case, that a certain statistical analysis of a deoxyribonucleic acid (DNA) match using the ceiling principle should have been excluded from evidence at the 1993 trial because the laboratory had not followed a certain calculation protocol, concluded that, in light of the repudiation in 1996 by the scientific community of the applicability of the ceiling principle to statistical analysis of DNA profile frequencies, the defendant had not been prejudiced by the admission of the statistical evidence, which was more favorable to him than that which would have been generated through the use of the scientifically accepted product rule: the jury verdicts of guilty were reinstated. [825-829]

INDICTMENTS found and returned in the Superior Court Department on April 15, 1993, and April 28, 1993, respectively.

The cases were tried before Walter E. Steele, J., and motions for required findings of not guilty and for reconsideration were considered by him.

Elspeth B. Cypher, Assistant District Attorney, for the Commonwealth.

Francis M. O'Boy (David M. Driscoll with him) for the defendant.

MARSHALL, J. A Superior Court jury convicted the defendant, Jeffrey Fowler, of rape of a child with the use of force and murder in the first degree (by reason of extreme atrocity or cruelty) the two year old daughter of the woman with whom he was living at the time. After sentencing, the trial judge, sua sponte, set aside the guilty verdicts and ordered a new trial. See Mass. R. Crim. P. 25 (b) (2), 378 Mass. 896 (1979). The Commonwealth appeals from that order and the judge's subsequent order denying the Commonwealth's motion for reconsideration. We vacate the order setting side the verdicts and ordering a new

Page 820

trial, with the result that the jury's verdicts of guilty of murder in the first degree and rape are to be reinstated.


The Commonwealth sought to introduce in evidence deoxyribonucleic acid (DNA) test results to establish that the defendant was the source of semen recovered from an autopsy swabbing of the victim's mouth. Forensic testing by the Federal Bureau of Investigation (FBI) using the Restriction Fragment Length Polymorphism (RFLP) method of analysis of DNA, [Note 1] concluded that DNA extracted from the semen from the victim's mouth matched DNA extracted from a sample of the defendant's blood.

Prior to trial, the defendant filed a motion in limine seeking to exclude any use of or reference to DNA evidence. The defendant did not challenge the RFLP analysis, [Note 2] but argued only that the statistical model used by the FBI to determine the probability that the DNA match might be random was not admissible under United States v. Frye, 293 F. 1013 (D.C. Cir. 1923), and Commonwealth v. Curnin, 409 Mass. 218 (1991). [Note 3] On November 8, 1993, the trial judge conducted a voir dire hearing

Page 821

to determine the admissibility of the DNA evidence. Supervisory Special Agent John Mertens of the FBI, appearing as an expert witness for the Commonwealth, described the DNA profiling process used by the FBI in this Case. Using a statistical model known as the product rule, [Note 4] Mertens estimated that the probability of the DNA sample obtained from the autopsy swabbing of the victim's mouth matching at random someone other than the defendant was one in 800,000. [Note 5]

The product rule now meets the test of scientific reliability stated in Commonwealth v. Lanigan, 419 Mass. 15 , 16 (1994) (Lanigan II). See Commonwealth v. Rosier, ante 807, 813-814 (1997). But its use has been -- and was at the time of the hearing -- a subject of controversy. In 1992, the National Research Council (NRC) [Note 6] published a report, entitled DNA Technology in Forensic Science (hereinafter the 1992 NRC Report), in which concerns about the product rule were noted: some scientists had concluded that genetic subpopulations might exist within a large genetic database population and might affect the statistical calculation used to determine the significance of a DNA match.

Page 822

To account for this possibility, the 1992 NRC Report recommended the use of a "ceiling frequency" for all DNA frequency calculations, known as the "ceiling principle." [Note 7] See generally Lanigan II, supra at 21. When used to calculate the probability of a DNA match in the general population, the ceiling principle was believed by some scientists to produce a more conservative estimate of the frequency with which a person's particular allele would appear in randomly selected population groups, and was thus thought to be more favorable to defendants than the product rule. See Lanigan II, supra at 23, and cases cited. Even at the time of the voir dire hearing in this case, however, the criticisms of the product rule were being questioned. The defendant's expert witness, Laurence D. Mueller, testified that some population geneticists had taken the position that the ceiling principle is not based on sound science and is unnecessary, while others had pointed out that the ceiling principle will not always result in a conservative estimate of a DNA profile frequency, and that in some cases it may result in a DNA profile frequency estimate that is more rare than it should be.

At the voir dire hearing Mertens testified that the FBI did not agree with the NRC about the existence of genetic subpopulations. Nevertheless, he said, the FBI had conducted an additional statistical analysis in this case using the ceiling principle. Calculated according to that principle, he said the chance of a random match of DNA from the autopsy swabbing from the victim and someone other than the defendant would be approximately one in 159,000.

Because the defendant's expert was not available at the time of the voir dire hearing, the judge made no ruling at that time; the evidence was closed "temporarily," subject to an expert's being called by the defendant. The trial commenced on November 29, 1993. On December 1, 1993, although the defendant had still called no DNA expert to testify, the judge again heard arguments from counsel concerning the admissibility of the Commonwealth's DNA evidence. The judge denied the defendant's motion in limine on December 6. On December 8, Mertens was called by the Commonwealth to testify. He told the jury that there was a match between the defendant's DNA

Page 823

and the DNA from the semen found in the victim's mouth. He also testified that, using the ceiling principle, the probability of the DNA sample obtained from the autopsy swabbing of the victim's mouth matching at random someone other than the defendant was approximately one in 150,000. Mertens's testimony before the jury included no discussion of any analysis using the product rule. [Note 8] The following day, with the agreement of the Commonwealth, the judge reopened the voir dire hearing, and the defendant called Laurence D. Mueller as an expert witness. He testified outside the presence of the jury.

The only basis on which Mueller sought to challenge the Commonwealth's DNA evidence concerned the FBI's purported failure to follow certain procedural steps in connection with its statistical analyses using the ceiling principle. [Note 9] He explained that the 1992 NRC Report contained several "permanent" and "interim" recommendations on statistical evaluations of DNA evidence using the ceiling principle, [Note 10] and that the procedure the FBI had followed to calculate the probability of a DNA random

Page 824

match using the interim ceiling principle departed from the recommendations made in the 1992 NRC Report. The FBI did not apply the "counting method," it failed to use all of the population databases available to it and, he said, the FBI did not follow NRC recommendations in the method the FBI used to estimate individual DNA band frequencies, called the "fixed bin" system; the NRC recommended the "floating bin" method. [Note 11] Using a larger sample of databases than those used by the FBI, and employing three different "bin" calculations, Dr. Mueller testified that he had conducted his own interim ceiling principle analysis of the DNA samples in this case, and that the likelihood of a random match was one in 136, one in 351, or one in 4,300. [Note 12]

A few days after Dr. Mueller had given his voir dire testimony, and as the trial progressed, the judge denied the defendant's attempt to exclude the testimony of Mertens that the jury previously had heard. Dr. Mueller never testified before the jury. On December 17, 1993, the jury convicted the defendant. The defendant promptly moved pursuant to rule 25 (b) (2) that the judge enter verdicts of not guilty, or order a new trial. The judge denied the motion. On January 7, 1994, the defendant was sentenced to life in prison pursuant to G. L. c. 265, s. 1, on the murder conviction, and to a concurrent life sentence on the rape conviction.

On February 11, 1994, the judge sua sponte issued a statement on the record in which he said that he had reconsidered the defendant's postconviction motion under rule 25 (b) (2), and

Page 825

that he would, as a matter of law, allow the motion, set aside the verdicts, and order a new trial. On March 31, 1994, the judge issued findings of fact and rulings of law in which he explained his decision; he concluded that the significance of the DNA match analyzed under the ceiling principle was not scientifically reliable and that the Commonwealth's DNA evidence should have been excluded. On November 23, 1994, the Commonwealth moved for reconsideration of the judge's decision, in light of Lanigan Il, [Note 13] which the motion judge denied on March 16, 1995. He ruled that although Lanigan II had determined that the ceiling principle was a reliable statistical methodology, the decision did not establish a per se rule as to the admissibility of the analysis in every case. Rather, the judge reasoned, his focus was "on whether the calculation procedure employed by the FBI in this case was a true and accurate execution of the ceiling principle." He concluded that the FBI's interim ceiling principle analysis was inadmissible because the FBI had failed to follow the recommendations in the 1992 NRC Report.


Under rule 25 (b) (2), [Note 14] a trial judge has the discretion to award a new trial on the ground that the verdict, although supported by legally sufficient evidence, is against the weight of

Page 826

the evidence, or because the integrity of the evidence is suspect. See Commonwealth v. Doucette, 408 Mass. 454 , 456 (1990), and cases cited. It is clear here that the judge's decision to set aside the verdicts and to award a new trial was grounded solely on his conclusion that the FBI's use of the interim ceiling principle to calculate the DNA profile frequency was not scientifically reliable. Because of developments that have occurred since his ruling, which we describe below, we conclude that in this case the judge's order must be vacated.

In Commonwealth v. Vao Sok, ante 787, 796-798 (1997), we today explain the standard we exercise when reviewing a trial judge's ruling on the admissibility of scientific evidence. [Note 15] First, when examining a trial judge's determination of the scientific validity of a particular methodology we said that "our review is de novo because a trial judge's conclusion will have applicability beyond the facts of the case before him." Id. at 797. Once we resolve the issue of the over-all scientific validity of the particular methodology, our review of the judge's rulings on admissibility "turns to traditional evidentiary inquiries." Id. at 798. At this stage "[t]he burden is on the proponent to show that the testifying expert properly performed a scientifically valid methodology in arriving at his opinion." Id. at 798, and cases cited.

Applying these standards of review, we turn to the judge's rulings concerning the admissibility of the Commonwealth's DNA evidence, in particular the scientific reliability of the statistical models used by the Commonwealth to describe the significance of the DNA match. This requires that we describe first a development in the relevant scientific community that occurred after the posttrial motions in this case. In 1996 the NRC issued a new report entitled, The Evaluation of Forensic DNA Evidence (1996 NRC Report), in which the NRC revised its earlier view on the ceiling principle analysis. The 1996 NRC Report at 5 recommended that in general the calculation of a

Page 827

DNA profile frequency should be made using the product rule. [Note 16] The NRC noted that it shared the view of those scientists who had criticized the ceiling principle "on practical and statistical grounds" and who see "no scientific justification" for the use of the ceiling principle. It agreed, it said, with the "wide[] criticisms" of the interim ceiling principle. [Note 17] The 1996 NRC Report concluded that "sufficient data have been gathered to establish that neither ceiling principle is needed." [Note 18] Id. at 35. The 1996 NRC Report also determined that the counting method was not required: "Since we believe that the abundant data make the ceiling principles unnecessary, this is true a fortiori for the direct counting method." Id. at 159. We conclude that neither the ceiling principle nor the interim ceiling principle is required for DNA profile frequencies.

The judge did not, of course, have the benefit of the 1996 NRC Report. In his 1994 ruling, made before our decision Lanigan 11, he concluded that the ceiling principle, and its application here, did not meet the requirements of scientific reliability. In his March, 1995, order denying the Commonwealth's motion to reconsider, he "assume[d] without deciding" that the ceiling and interim ceiling principles were scientifically valid, but concluded that the FBI's application of the interim ceiling principle was unreliable. It is now clear that the concerns that the judge had, both with respect to the interim ceiling principle

Page 828

generally and as applied by the FBI in this case, are no longer valid.

As described above, the 1996 NRC Report repudiated the need for the ceiling principle analysis, and any possible failings of the FBI in executing the ceiling principle (and, in light of that Report, we are not persuaded that there were any such failings) are also laid to rest. [Note 19] Moreover, there is nothing on this record to suggest that the FBI's analysis using the product rule to calculate the DNA profile frequency was invalid. In these circumstances we conclude that there is no prejudice to the defendant by the admission of the Commonwealth's DNA profile frequency evidence. [Note 20] The jury convicted the defendant in the face of the FBI expert's testimony at trial that the likeli-

Page 829

hood of the DNA match with the defendant occurring at random was one in 150,000. The FBI expert testified at the voir dire hearing (but not at trial) that the probability of a random match using the product rule was one in 800,000. That result, if introduced in evidence, would have been far more prejudicial to the defendant than the results of the FBI's analysis using the ceiling principle. [Note 21]

In our two-pronged review of the admission of scientific evidence, we afford great deference to the decisions of trial judges concerning the reliability of the application of a scientific technique in a particular case. But in this case all of the judge's concerns stem from challenges to the ceiling principle that have since been resolved by the 1996 NRC Report and related studies.

The order setting aside the defendant's verdicts and granting the defendant a new trial is vacated and the jury's verdicts of murder in the first degree and rape are reinstated. Sentence is to be imposed in the Superior Court.

So ordered.


[Note 1] There are different methods used in DNA forensic testing including RFLP and polymerase chain reaction (PCR) amplification and allele identification. See Commonwealth v. Vao Sok, ante 787, 789-792 (1997). RFLP analysis identifies and analyzes genetic loci containing different numbers of copies of genetic base sequences, called "variable number of tandem repeats" (VNTRs). See Commonwealth v. Curnin, 409 Mass. 218 , 227-230 (1991). For a comparison of the RFLP and PCR-based methods of DNA forensic testing, see Vao Sok, supra at 790-792. The PCR-based method of analysis is not at issue in this case.

[Note 2] The RFLP method of analysis has been found to be scientifically reliable. Commonwealth v. Lanigan, 419 Mass. 15 , 27 (1994) (Lanigan II). Commonwealth v. Daggett, 416 Mass. 347 , 350 n.1 (1993).

[Note 3] Under United States v. Frye, 293 F. 1013 (D.C. Cir. 1923), a party offering evidence produced by a scientific theory or process had "the burden of showing the general acceptance by experts in the field of the reliability" of that evidence. Commonwealth v. Kater, 388 Mass. 519 , 527 (1983). In Commonwealth v. Curnin, supra at 227, applying the Frye test, we held that the admission in evidence of DNA test results constituted prejudicial error where there was an absence of general acceptance or inherent rationality of the process by which the testing laboratory arrived at the probability of the DNA match occurring in the population at large. evidence is relevant and reliable. In Lanigan II, supra at 26, we acknowledged Daubert, and described the standard of admissibility for scientific evidence in our courts:

"We accept the basic reasoning of the Daubert opinion because it is consistent with our test of demonstrated reliability. We suspect that general acceptance in the relevant scientific community will continue to be the significant, and often the only, issue. We accept the idea, however, that a proponent of scientific opinion evidence may demonstrate the reliability or validity of the underlying scientific theory or process by some other means, that is, without establishing general acceptance."

See Commonwealth v. Sands, 424 Mass. 184 , 185-186 (1997).

In 1993, the Supreme Court decided Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), in which it held that Fed. R. Evid. 702 superseded the Frye test. Id. at 588. Rule 702 imposes a duty on the trial judge to make an initial determination whether scientific testimony and

[Note 4] We discussed the product rule in Commonwealth v. Curnin, supra at 224, and in Commonwealth v. Rosier, ante 807, 813 (1997). In brief, "[u]nder the product rule, the frequency in the population base of each allele disclosed in the DNA test is multiplied to produce the frequency of the combination of all the alleles found." Lanigan 11, supra at 21, citing Commonwealth v. Curnin, supra at 224 n.10, 225 n.11.

[Note 5] Although Mertens did not use the precise term, the procedure he described at the voir dire hearing for calculating the DNA match frequency statistic was the product rule model, as the judge noted in his order granting the defendant a new trial.

[Note 6] The National Research Council is an association of leading scientists, the members of which are drawn from the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine.

[Note 7] For a discussion of the ceiling principle, see Commonwealth v. Lanigan, 413 Mass. 154 , 163 (1992) (Lanigan I); Commonwealth v. Daggett, supra at 356-357 n.1 (Abrams, J., concurring); Lanigan II, 419 Mass. 15 , 27 (1994); Commonwealth v. Rosier, supra at 814-815.

[Note 8] Although the record does not reflect any ruling by the judge that Mertens could not testify before the jury about the FBI's analysis of the DNA samples using the product rule, the record suggests that the judge was greatly concerned about the admissibility of testimony concerning the product rule in light of the dispute concerning its reliability within the scientific community at that time.

[Note 9] As noted above, the defendant never challenged the RFLP method of analyses used by the FBI, and he never suggested that any aspect of the FBI's DNA profiling -- other than the ceiling principle statistical analysis -- was scientifically unreliable or invalid in any respect. As the judge noted in his March, 1995, order denying the Commonwealth's motion for reconsideration, "the only part of the DNA evidence challenged by the defendant under the Frye-Curnin standard, was the statistical procedure, known as the 'ceiling' method, used to calculate the significance of a laboratory match of DNA profiles. "

[Note 10] Mueller testified that the 1992 NRC Report contained three permanent recommendations: (1) that all forensic laboratories base their statistical analyses on a collection of fifteen to twenty ethnic subgroup database populations rather than to use three or four different racial categories; (2) that the product rule be replaced by the ceiling principle for calculating the probability of a DNA match in the general population; and (3) that laboratories report to juries the false positive rate in order to indicate the probability that the laboratory might incorrectly match DNA samples.

Mueller further testified that the 1992 NRC Report had recognized that it would take some time for laboratories to collect samples for the fifteen to twenty ethnic subgroup database populations, and had therefore made two "interim" recommendations for laboratories to follow while they assembled the ethnic subgroup database populations. These interim recommendations, Mueller said, are (1) that laboratories use a "counting method," by which the DNA profile obtained from the evidence sample be compared to every person in the laboratory's database to see whether there is any match, and (2) that the laboratories perform a "modified" ceiling principle, using at least three databases. Mueller noted that the 1992 NRC Report also recommended that during the interim period, laboratories report their laboratory error rates.

[Note 11] In determining whether a DNA fragment from one sample matches another DNA fragment, the fragments' sizes are compared. The range within which DNA fragments of different sizes cannot be reliably distinguished is called a "bin." DNA fragments within a bin are treated as though they are the same size. See S. Easteal, N. McLeod & K. Reed, DNA Profiling: Principles, Pitfalls and Potential 193 (1991). Under the fixed bin analysis, the boundaries of a bin are determined by the position of DNA fragments of known size. Under a floating bin analysis, the bins, boundaries are determined by making repeated measurements of the same DNA fragments run on different gels used in the separation of the fragments. Id. at 92.

[Note 12] Dr. Mueller used fixed bins, floating bins with a 5% statistical interval, and floating bins with a 9% statistical interval, respectively.

[Note 13] In Lanigan II, supra at 26-27, we upheld the admission at Lanigan's trial of evidence of the significance of a DNA match, calculated according to the ceiling principle.

[Note 14] Rule 25 of the Massachusetts Rules of Criminal Procedure, 378 Mass. 986 (1979), provides in relevant part:

"(a) ENTRY BY COURT. The judge on motion of a defendant or on his own motion shall enter a finding of not guilty of the offense charged in an indictment . . . after the evidence on either side is closed if the evidence is insufficient as a matter of law to sustain a conviction on the charge. . . .

"(b) (2) MOTION AFTER DISCHARGE OF JURY. If the motion is denied and the case is submitted to the jury, the motion may be renewed within five days after the jury is discharged and may include in the alternative a motion for a new trial. If a verdict of guilty is returned, the judge may on motion set aside the verdict and order a new trial, or order the entry of a finding of not guilty, or order the entry of a finding of guilty of any offense included in the offense charged in the indictment or complaint."

[Note 15] That standard of review is relevant here. The judge emphasized that he was allowing the defendant's postconviction rule 25 (b) (2) motion for a required finding of not guilty, not as an exercise of discretion, but as a matter of law, because "there was no general acceptance within the relevant scientific community of the ceiling method or interim ceiling method as a reliable statistical calculation, [and] as a matter of law, neither the evidence of the laboratory DNA 'match', nor the statistical evidence interpreting that 'match', should have been admitted at trial under the Frye-Curnin standard."

[Note 16] In Rosier, supra at 817, we said that "[w]e agree with the conclusions of the 1996 NRC Report," and recognized the scientific validity of the product rule in the context of the PCR-based method of analysis of DNA samples. The product rule has validity as well in the context of the RFLP method of analysis used in this case. The 1996 NRC Report recognizes that the ceiling principle, developed in response to concerns about the possibility of genetic subpopulations, was of particular significance in calculating VNTR profile frequencies, such as the profile frequencies in this case. See note 1, supra. The 1996 NRC Report's conclusion, supra at 158, that the ceiling principle and interim ceiling principle were "excessively conservative," and its conclusion that the product rule should be used in making profile frequencies is applicable to VNTR-based systems, such as the RFLP. Id. at 5.

[Note 17] In anticipation that some laboratories might still use the ceiling principle analysis, the 1996 NRC Report made recommendations "that will make the [ceiling] principle more workable and less susceptible to creative misapplications."

[Note 18] A 1993 report produced by the FBI, entitled VNTR Population Data: A Worldwide Study arrived at a similar conclusion that the product rule was valid and that the data do not support the need for alternative procedures, such as the ceiling principle. See Rosier, supra at 815 n.16.

[Note 19] The judge identified four "failings" of the FBI in its analysis using the interim ceiling principle, none of which withstands scrutiny in light of the 1996 NRC Report. The judge found that "the FBI did not apply the counting method statistical calculation and instead used a corrective mathematical ceiling approach for which there was no support in the relevant scientific community." The 1996 NRC Report discredits any requirement that the "counting method" be used. Id. at 159-160. Second, the judge said "the FBI failed to use all of the databases at its disposal." The 1996 NRC Report validates the use by the FBI of the population groups in this case. Third, the judge concluded that, "although the NRC did not recommend that a laboratory's error rate in conducting an interim ceiling calculation should be told to the jury, as the NRC stated should be done for a standard ceiling test, Dr. Mueller stated that such an error-rate report should also be made to the jury for any interim ceiling tests." The 1996 NRC Report, at 185, explicitly declined to make such a recommendation, commenting that "[i]nasmuch as the purpose of our report is to determine what aspects of the procedures used in connection with forensic DNA testing are scientifically valid, we attempt no such policy judgment" (emphasis supplied). Finally, the judge noted that "although the FBI used the 'fixed-bin' method instead of the 'floating-bin method' as recommended by the NRC, this is not fatal to the FBI's calculation because the NRC also stated that fixed-bin method was 'acceptable' for purposes of the interim ceiling approach." This is consistent with the 1996 NRC Report at 162.

[Note 20] The lack of prejudicial error resulting from admitting the Commonwealth's evidence of the DNA profile frequency is further evidenced by the limited range of men who could have been the source of the semen found in the victim's mouth. It is undisputed that the victim, a two year old child, had contact with few people (all identified) in the days immediately preceding her death. The Commonwealth accounted for the last days of the victim's life, and tested the DNA of the only four males who had contact with the victim in the days immediately preceding her death: the defendant; the victim's father; a friend of the father; and the victim's maternal grandfather. The Commonwealth's expert testified that all but the defendant were eliminated as the source of the semen found in the victim's mouth.

[Note 21] As Mertens testified, the FBI studies indicated that genetic subpopulations did not exist and therefore the ceiling principle analysis was not needed. But to the extent there was any doubt about that, the ceiling principle eliminated any effect that the subpopulations might have and added "further level of conservativeness to an already conservative system" to the statistical analyses using the product rule. The defendant's expert agreed.