In November, 1992, a jury convicted the *817defendant of murder in the first degree of Richard M. Alfredo.1 The appeal was entered here in January, 1995. On October 18, 1995, the defendant filed a motion for a new trial which a single justice of this court ordered transferred to the Superior Court for disposition. The trial judge had retired, and another judge held an evidentiary hearing on the motion in May, 1996.
On June 30, 1997, based on his conclusions stated in a thorough memorandum which considered the evidence at trial and the evidence presented to him, the judge allowed the new trial motion. He did so because of the “effect of the Commonwealth’s failure to timely disclose . . . evidence [to defense counsel] combined with defense counsel’s failure to present a competent rebuttal of the prosecution’s case for cause-of-death.” The Commonwealth’s appeal from the allowance of the motion for a new trial is now before us.
The Commonwealth’s theory at trial was that the defendant put LSD (lysergic acid diethylamide) in food served to Alfredo, who had severe heart disease, and that he died as a result of heart disease and acute LSD intoxication, which caused an abnormal, fast heart beat. The motion judge determined that the defense counsel had been ineffective, in a constitutional sense, in not adequately challenging the Commonwealth’s evidence that LSD was found during post mortem tests of Alfredo and that the Commonwealth failed adequately to disclose test results that did not confirm, indeed in part tended to disprove, the presence of LSD in Alfredo’s body.
Generally, we consider whether a motion judge committed a significant error of law or other abuse of discretion in allowing a defendant’s motion for a new trial. Commonwealth v. Grace, 397 Mass. 303, 307 (1986). By that standard, the issue in this case is whether the judge erred in ruling, on his largely unchallenged findings, that the defendant’s trial counsel was ineffective, that the Commonwealth prejudicially failed to provide specific information that it was directed to provide, and that consequentially the defendant was entitled to a new trial.2
We summarize facts that the jury could have found, with *818particular emphasis on the evidence concerning whether LSD could have been a factor in Alfredo’s death. The defendant and Alfredo, who was sixty-one years old at his death, had been living together in Westport for many years. There was evidence that she intended to kill Alfredo. The defendant knew that Alfredo suffered from coronary artery disease. She thought drugs would kill him, given his health condition. She purchased what she thought were drugs from local teenagers and told them that she intended to use them to kill her boy friend. Showing what the motion judge described as “common sense and an enterprising spirit,” the teenagers sold her bogus substitutes. Ultimately, however, the defendant apparently succeeded in purchasing real drugs. She thought she had purchased mescaline, although she may in fact have purchased LSD, which the evidence indicated is often sold as mescaline.
The Commonwealth’s theory was that the defendant caused Alfredo’s death by putting LSD in his jello. The defendant told relatives and others that she had put mescaline in the jello and had given it to the deceased. The defendant’s daughter Teasha, who was fourteen at the time of Alfredo’s death and almost eighteen at the time of trial, testified as a defense witness that it was she who had put the drug in the jello and had given it to Alfredo. At the last minute, according to Teasha and the defendant, the defendant backed out of the plan, and Teasha *819alone administered the substance to Alfredo. There was testimony from both Teasha and the defendant that Alfredo died shortly after eating the jello. Teasha testified that Alfredo had sexually abused her.
A doctor concluded that Alfredo had died of an apparent myocardial infarction. After suspicion had focused on the defendant and thirty-one days after Alfredo’s death, his body was exhumed. The body had been embalmed. Water had entered his casket.
The scientific evidence at trial of the cause of death was as follows. A pathologist testified that he performed an autopsy on the exhumed body. Alfredo had severe heart disease and had had a coronary bypass procedure. His heart was enlarged. He had had at least one heart attack more than six months earlier but was not having one when he died. The pathologist took samples of body fluids that had apparently not mixed with embalming fluid. He submitted half the samples to the State police toxicology laboratory and half to a private laboratory. He testified, without objection, that the laboratory reports indicated the presence of LSD in all of the samples. He testified that Alfredo died as a result of a combination of heart disease and acute LSD intoxication that caused increased stress on the diseased heart.
On cross-examination, the pathologist testified that he was “comfortable” with his opinion on the cause of death, although he had never before seen LSD cause death, and that no one has established the toxic level of LSD. Defense counsel did not ask what tests had been conducted, nor did he establish that, as the motion judge found, substances other than LSD could have produced positive test results in the screening tests that the pathologist conducted.3
A clinical chemist, part owner of the private laboratory to *820which samples had been sent, testified next. Tests of the samples that the laboratory received did not disclose mescaline but did disclose a compound found in cold relief medicine, which itself can cause an increased heart beat. The laboratory next used a radioimmunoassay (RIA) to test for LSD, because it is often sold as mescaline. The chemist said that all three samples tested were positive for LSD. The witness testified further that, on receipt of a positive result in a RIA test, one should conduct a gas chromatography mass spectrometry (GCMS) test. He added, without objection, that samples were sent to another laboratory for that test and its reports confirmed the presence of LSD.4
A chemist employed by the Massachusetts State police crime laboratory testified that he had tested the samples for mescaline and found none. He then used an RIA test to screen certain samples for LSD and obtained a positive result. He then testified that: “I attempted to confirm the presence of LSD using a separate different test[] known as gas chromatography mass spectrometry on the autopsy samples. I was not able to confirm the initial screening test that I obtained using radioimmunoassay.” On cross-examination, the chemist testified that, in his experience, a positive RIA test result “indicates the real presence of LSD in the sample.” He then added incorrectly that the failure to confirm the presence of LSD using GCMS indicated simply that one test was positive and one was negative.5
Next, a cardiologist testified, without objection, that the toxicologic examination “showed the presence of LSD on positive screens in all of the fluids and organs that were examined.” He concluded that “the combined effects of the LSD in [Alfredo] acting upon his severe heart disease caused his death” because of a cardiac arrest. On cross-examination, he testified that he had never seen a death caused by LSD poisoning.
*821The motion judge heard testimony on the methodology, accuracy, and validity of different tests which detect the presence of a chemical compound. He concluded that RIA is a screening test that indicates the possible presence of a compound but needs a confirmatory test, such as GCMS, to substantiate its result. See Greenblatt, Urine Drug Testing: What Does it Test? 23 New Eng. L. Rev. 651, 655 n.25, 656-657 (1988).
The motion judge concluded that the defendant was entitled to a new trial. He did so in part because defense counsel failed to deal properly with the Commonwealth’s claim that LSD was a cause of Alfredo’s death. The judge found that the cause of death had been barely contested at trial. Defense counsel had obtained funds to hire an expert, had consulted a well-known toxicologist, but presented no expert testimony. Defense counsel could have used that expert to reveal defects in the scientific evidence.6 The motion judge concluded “that defense counsel did not appreciate that only a screening test had been offered in evidence which was of no probative value without a confirmatory test. The concept of a ‘confirmatory test’ was completely missed at trial. The fact that the screening test had no value without a confirmatory test did not register in any of trial counsel’s statements or actions.” He added that “it is not conceptually difficult to distinguish between a test which screens for a substance” and one that is needed to confirm a preliminary indication. The judge accepted “that the standard, conventional practice recognized as most scientifically accurate among toxicologists is to report as negative any RIA test that was not confirmed by a more accurate confirmatory test,” because an inability to confirm an initially positive RIA test “indicates the presence of a cross-reactant rather than the target drug.” See Miike, Accuracy and Reliability of Urine Drug Tests, 36 Kan. L. Rev. 641, 646-648 (1988) (“Confirmatory tests are necessary to distinguish between positive results due to the drug’s presence . . . [and] positive results due to cross reactivity . . .”).
The cause of death issue was not fully presented to the jury. The Commonwealth was permitted to present its case as if the screening test results had established that LSD was in Alfredo’s body. The only direct evidence at trial of the results of a satisfactorily conducted confirmatory test was that the GCMS *822test that the chemist from the State police crime laboratory ran did not confirm the presence of LSD.7
The motion judge considered whether the failure to pursue the cause of death question might have been a defensible tactical decision. An attorney’s tactical decision amounts to ineffective assistance of counsel only if it was manifestly unreasonable when made. Commonwealth v. Roberts, 423 Mass. 17, 20 (1996). Commonwealth v. Adams, 374 Mass. 722, 728 (1978). There was no inhibiting conflict between arguing the cause of death issue and presenting the testimony of the defendant’s daughter Teasha that she, acting alone, had put a drug in Alfredo’s food. This case does not involve a reasonable tactical decision. The force of Teasha’s confession was weakened by the fact that she was effectively beyond prosecution for her alleged actions while a juvenile.
We agree with the motion judge that defense counsel was ineffective in a constitutional sense. His failure to challenge the evidence that Alfredo’s body had LSD in it was an error that falls below acceptable constitutional standards. See Commonwealth v. Haggerty, 400 Mass. 437, 442-443 (1987). Certainly evidence and cross-examination that challenged the presence of LSD in Alfredo’s body could have raised a reasonable doubt in the minds of the jury. The defendant was thus deprived of “an otherwise available, substantial ground of defence” (Commonwealth v. Saferian, 366 Mass. 89, 96 [1974]) or, otherwise stated, “better work might have accomplished something material for the defense” (Commonwealth v. Satterfield, 373 Mass. 109, 115 [1977]). If defense counsel had presented evidence that reliable tests had failed to confirm the presence of LSD in Alfredo, or even that no reliable confirmatory tests had been conducted, the defendant would have had a strong argument that Alfredo did not ingest LSD. Any opinion that LSD was a cause of Alfredo’s death could then have been seriously undercut.
The judge allowed the motion for a new trial, in part because the Commonwealth failed to produce exculpatory evidence that the defendant had specifically sought by a pretrial motion to *823which the Commonwealth assented and a magistrate allowed. The defendant requested a description of all scientific tests, the manner in which they were performed and by whom, and the results of each test. The prosecution turned over all the relevant test information in its possession. The prosecution did not, however, turn over information, not in its actual possession, concerning five inconclusive or failed confirmatory tests conducted by the State police laboratory chemist. See note 7, supra.
The Commonwealth had the obligation to produce all test results in its possession. Its failure to do so would require a new trial if admission of the test results could have been a real factor in the jury’s deliberations. See United States v. Agurs, 427 U.S. 97, 104 (1976); Commonwealth v. Tucceri, 412 Mass. 401, 414 (1992). We have already discussed the significance of a failed attempt to use GCMS to confirm the presence of LSD in a sample that RIA testing had showed contained LSD or a cross reactant. The disclosure of these test results before trial might have alerted defense counsel to the possible weakness of the Commonwealth’s attempt to prove that LSD was in Alfredo’s body. Defense counsel’s omissions, in any event, do not relieve the Commonwealth of the consequences of its failure to disclose. Commonwealth v. Tucceri, supra at 410. Certainly, if the undisclosed test results had been admitted in evidence and properly interpreted by an expert, the defendant would have been substantially aided in seeking to create a reasonable doubt that LSD was a cause of Alfredo’s death. See Commonwealth v. Cronk, 396 Mass. 194, 200 (1985).
The question remains whether the Commonwealth, subject to a specific direction to produce evidence, may be exonerated from failing to produce State police laboratory test results because that evidence was not in the prosecution’s possession (or known to it), but rather was held by the State police crime laboratory and perhaps known only to one of its chemists.
The prosecution had a duty to inquire concerning the existence of scientific tests, at least those conducted by the Commonwealth’s own crime laboratory. That obligation is inherent in the allowance of the motion to produce all scientific test results. The prosecutor’s office obviously did not conduct scientific tests. It could not satisfy the production order simply by turning over test information that it had in its files. It had a duty of inquiry. See Kyles v. Whitley, 514 U.S. 419, 438 (1995). *824A prosecutor’s obligations extend to information in possession of a person who has participated in the investigation or evaluation of the case and has reported to the prosecutor’s office concerning the case. Commonwealth v. Daye, 411 Mass. 719, 733 (1992). Commonwealth v. St. Germain, 381 Mass. 256, 261 n.8 (1980). Such a person is sufficiently subject to the prosecutor’s control that the duty to disclose applies to information in that person’s possession. See Kyles v. Whitley, supra at 437 (“the individual prosecutor has a duty to learn of any favorable evidence known to the others acting on the government’s behalf in the case, including the police”); Commonwealth v. Gallarelli, 399 Mass. 17, 20 n.4 (1987) (“The Commonwealth does not dispute that, as matter of law, the [Boston police crime] laboratory report was within its custody and control, even though no copy was ever in the prosecutor’s files, and he himself was not aware that it existed. See Commonwealth v. St. Germain, 381 Mass. 256, 261 n.8 [1980]. The Commonwealth also concedes that the prosecutor was duty bound to disclose the contents of the laboratory report”). The Commonwealth should have disclosed before trial the fact and the results of the attempted confirmatory tests conducted by the State police laboratory.
We affirm the motion judge’s order granting the defendant a new trial.
So ordered.
A transcript was promptly ordered, and a notice of appeal filed. Almost two years later the transcript was filed, and approximately three months later, trial counsel having withdrawn, appellate counsel was assigned.
If we were to disagree with the motion judge and reject his grant of a new trial, we would in due time, pursuant to G. L. c. 278, § 33E, have to decide the defendant’s appeal from her conviction of murder in the first degree. In *818doing so, we would consider, among other issues, whether a new trial was needed in this so-called “capital case” because one or more errors created a substantial likelihood of a miscarriage of justice. G. L. c. 278, § 33E. This standard of appellate review is more favorable to a defendant than is the standard applied to a claim of ineffective assistance of counsel. Commonwealth v. Wright, 411 Mass. 678, 682 (1992).
An error creates a substantial likelihood of a miscarriage of justice if it was likely to have influenced the result. Id. This test, applicable in a capital case, devotes no attention to the question, inherent in any ineffective assistance of counsel claim (see Commonwealth v. Satterfield, 373 Mass. 109, 115 & n.10 [1977]), whether counsel’s behavior fell measurably below what is expected of an ordinary fallible lawyer. Commonwealth v. Parker, 420 Mass. 242, 245-246 (1995), and cases cited.
The defendant has not argued that we should consider the questions before us on the standard of review applicable under § 33E. It is not established that the denial of a motion for a new trial in a capital case, appealed in conjunction with the defendant’s direct appeal, is to be reviewed pursuant to § 33E. See Commonwealth v. Finstein, 426 Mass. 200, 204 (1997); Commonwealth v. Robinson, 382 Mass. 189, 209 (1981). Because we agree with the motion judge’s rulings on the theories argued to him (and us), we need not pursue the point.
After the pathologist had testified, the trial judge had the following discussion with defense counsel:
The judge: “Let me ask you . . . with respect to the examination of the doctor and some of the other witnesses, they haven’t been cross-examined that much in detail. Is that part of your strategy?”
Defense counsel: “It’s part of the defense that I intend to put on next week, your Honor. It’s unnecessary to a great extent of detail — to try to resolve little issues would be a mistake.”
The judge: “That’s the way your client wants you to proceed?”
Defense counsel: “Yes.”
The evidence at trial did not identify this laboratory, nor were its test results admitted in evidence.
That statement was inaccurate and seriously misleading. The failure to confirm the presence of LSD made the positive result in the screening tests of low probative value. See Imwinkelried, Forensic Science: Immunoassay Drag Testing, 28 Crim.'L. Bull. 158, 163 (1992) (“False positives are common . . . with error rates exceeding 60 percent”). Gas chromatography mass spectrometry “is, for all practical purposes, 100% accurate.” Pella v. Adams, 702 F. Supp. 244, 246 (D. Nev. 1988).
The positive RIA test could have been caused by a “cross reactant,” such as ergot fungus which can be present in grass and soil and might have been in the groundwater that seeped into the casket. Formaldehyde, which is an embalming fluid, is itself a cross reactant.
Before trial, defense counsel wrote the toxicologist to ask for his assistance “on what is probably the seminal issue in this case.”
We know from the evidence presented at the hearing on the motion for a new trial that the chemist made five attempts. In some tests he obtained an inconclusive result and in at least one he found no LSD. The Commonwealth had not disclosed the fact of these five tests to defense counsel before trial, a point to which we shall return.