Gilday v. Callahan

COFFIN, Senior Circuit Judge. Petitioner William Gilday was convicted of first degree murder and two counts of armed robbery for his involvement 25 years ago in a notorious bank robbery in which Boston Police Officer Walter A. Schroeder was killed. This habeas case, originally filed in 1981, was reactivated after disposition of the last of his four unsuccessful motions for new trial in the Massachusetts courts. The district court denied the petition in a comprehensive opinion. 866 F.Supp. 611 (D.Mass.1994). After carefully reviewing the case authorities and relevant portions of the record, we affirm. *260 I. Background We shall provide at this juncture only brief factual background, adding more details in later sections as necessary for an understanding of the issues discussed. A lengthy description of the evidence presented at Gil-day’s five-week trial is reported in Commonwealth v. Gilday, 367 Mass. 474, 478-485, 327 N.E.2d 851, 854-58 (1975) (“Gilday I”). See also Gilday, 866 F.Supp. at 640-43. A full chronology of the proceedings since his 1972 conviction is set out in the district court’s opinion. Id. at 615-16. Gilday and five others were indicted on robbery and murder charges. 1 Evidence indicated that the group had planned a series of bank robberies to raise funds in support of radical political activities. The Supreme Judicial Court summarized as follows the evidence supporting the Commonwealth’s theory of what occurred on the day of the robbery at issue here: Bond, Valeri and Saxe entered the bank carrying guns, robbed it and drove off in a blue Chevrolet---- Gilday, armed with a semiautomatic rifle, was seated in a white Ambassador automobile across the street from the bank____ [Ajfter the other three had escaped from the scene, Gilday fired a number of shots at two policemen who arrived, and Officer Sehroeder thereby sustained the wounds from which he died the next day. Bond, Valeri, and Saxe later switched to a third vehicle, a station wagon driven by Power, and made their escape. Gilday also escaped in the white Ambassador. 367 Mass. at 477, 327 N.E.2d 851. On March 10, 1972, Gilday was convicted by a jury and sentenced to death. Following the United States Supreme Court’s decision in Furman v. Georgia, 408 U.S. 238, 92 S.Ct. 2726, 33 L.Ed.2d 346 (1972), and his first motion for new trial, the death sentence was changed to a sentence of life imprisonment. His subsequent efforts to obtain relief from the original convictions have proven unsuccessful. In this appeal, Gilday argues that he is entitled to a writ of habeas corpus because his trial was replete with constitutional error, and there consequently is substantial reason to believe he was innocent of the charges on which he was convicted. We have considered each of his claims fully, but cannot say that any of the identifiable flaws in the proceedings constituted a deprivation of rights warranting reversal of his convictions. We discuss most of these claims in some detail below. As for the others, the district court’s analysis so closely reflects our own thoughts that we find it unnecessary to repeat the discussion and, therefore, adopt its conclusions as our own. II. Reasonable Doubt Instruction Gilday claims a host of problems with the trial judge’s reasonable doubt instruction, several of which center on language that has been expressly and repeatedly disapproved by this and other courts. Because we agree that this charge was flawed, we have studied its full text and context with particular care in order to answer the relevant constitutional question: “whether there is a reasonable likelihood that the jury understood the instructions to allow conviction based on proof insufficient to meet the [reasonable doubt] standard,” Victor v. Nebraska, — U.S. -, -, 114 S.Ct. 1239, 1243, 127 L.Ed.2d 583 (1994). Our review is de novo. See Ouimette v. Moran, 942 F.2d 1, 4 (1991) (presumption of correctness for state court findings of fact under 28 U.S.C. § 2254 applies only to ‘“basic, primary or historic facts’” (citation omitted)). As we previously remarked when evaluating a strikingly similar instruction in Bumpus v. Gunter, 635 F.2d 907, 910 (1st Cir. *261 1980), 2 “[i]t is to be remembered ... that [the challenged] remarks have been separately culled from a very lengthy charge. They, and the emanations from them, must be assessed along with the rest of the charge....” The Supreme Court recently reaffirmed the need to examine a charge in context to determine whether language possibly erroneous in the abstract is cleansed because “the rest of the instruction ... lends content to the phrase,” Victor, — U.S. at -, -, 114 S.Ct. at 1247, 1250-51. See also id. at -, 114 S.Ct. at 1243 (“ ‘[T]aken as a whole, the instructions [must] correctly conve[y] the concept of reasonable doubt to the jury.’” (quoting Holland v. United States, 348 U.S. 121, 140, 75 S.Ct. 127, 138, 99 L.Ed. 150 (1954))). In the end, we have come to the conclusion that the charge overall left the jury with an accurate impression of the substantial burden faced by the prosecution in establishing the defendant’s guilt beyond a reasonable doubt. As shall become apparent from our discussion below, none of the problems identified by Gilday is, on its own, of a severity that warrants reversal of his conviction. Indeed, several of the flaws are significantly ameliorated by other aspects of the charge. And, while the cumulative impact of the flaws is itself a separate matter of concern, we are persuaded that it does not rise to the level of constitutional error. The charge, which spanned 20 paragraphs when reduced to writing, 3 loosely may be divided into three separate segments for purposes of our review. We therefore begin our discussion by reproducing the first five paragraphs: It is the burden of the Commonwealth to establish its case beyond a reasonable doubt. What do those words mean? Well, you give to them their common ordinary meaning. A doubt means an uncertainty of mind or a lack of conviction. And reasonable means based upon a reason. I am going to discuss with you what our Court has said it does not mean. It does not mean a whimsical or a fanciful doubt; that is, a doubt which is conjured up, which has no strength to tie it together, which has no foundation in fact. It is floating around in the air. And you can’t pull it down and root it to something solid in the evidence. It is whimsical. It is not beyond all doubt. There are few things in this world of ours which are capable of proof beyond all doubt. That is an impossible burden. And if that were the burden that we placed upon the Commonwealth, no one who transgressed the laws of society or outraged our populace would ever be convicted of a crime. Don’t confuse beyond a reasonable doubt with beyond all doubt. And I sometimes think the jurors take that as their standard. They must be satisfied before they find a defendant guilty that there is no possibility that they are wrong before their full conviction. And so again, the Court has said: “Proof beyond a reasonable doubt is not beyond the possibility of innocence,” because I suppose almost anything is possible. And if you are satisfied as I define reasonable doubt of the proof of the Commonwealth’s case beyond a reasonable doubt you should not hesitate because of a haunting thought that there is a possibility that you might be wrong. Because then you place on the shoulders of the sovereign state a burden it does not have. The judge thus began simply, telling the jurors that a reasonable doubt is an uncertainty “based upon a reason.” Petitioner argues that the charge contained such a catalogue of examples of what was not reasonable doubt that the jury was in effect improperly influenced to assign whatever doubt it had to these examples. But tautology is not multiplicity; all of the references carried the identical message: that proof beyond a reasonable doubt is not beyond all *262 doubt. As we said in Bumpus, 635 F.2d at 911: While the judge placed what we regard as an uncomfortable degree of emphasis on the limits of the government’s burden, ... the charge in its entirety was not so unbalanced as to undercut the reasonable doubt standard, nor was it basically inaccurate. The next six paragraphs contain all of the troubling language. The section begins with a rhetorical question: “[s]o what does it [reasonable doubt] mean?” The judge then answered: Not one who is searching for a doubt to acquit; not one who has made up his mind that the defendant is not guilty, and then having decided the ultimate question, to satisfy his conscience goes back through the evidence and pores through it to find something upon which to pin the doubt which he already has. No, indeed. It is the doubt of a conscientious juror who is earnestly seeking the truth in the fullest discharge of the oath that he took. It is proof, as our Supreme Judicial Court has said, “To a moral certainty.” That is not a mathematical certainty; that is not a scientific certainty which is capable of exactness, because human beings are endowed with a free will; and they are capable of independent action. And you can’t take their conduct and put it into a test tube or a computer and come out with a nice answer. When you get all through analyzing this evidence, it has to be a doubt nagging your mind, leaving you with an uncertainty of conviction to that moral certainty which you can stand up and argue in the jury room with principle and integrity and honesty to your fellow jurors. And if you don’t believe in it yourself, you haven’t got a reasonable doubt. The Supreme Court has expressed it as, “The same degree of satisfaction of mind and conscience that jurors should have when they take action in the major affairs of their lives,” the major affairs of their lives. I do not attempt to define for you what are the major affairs of your lives. I leave it to your experience and I leave it to your wisdom. When you take vital action in your everyday lives certainly you should be satisfied to a moral certainty that what you are doing is right. None of us have a crystal ball. The future is not ours to see. All we can do is weigh the pros and the cons against any contemplated course of action; and then with the wisdom and the intellect that we possess, make a decision. We may be right; we may be wrong. But if we are satisfied to a moral certainty when we do an act in our private lives, that it is the right thing to do, we have a settled conviction of mind. That is the degree of proof which the law contemplates when they talk about “proof to a moral certainty.” Petitioner directs his fire to four problem areas in these passages: (1) the use of the term “moral certainty,” (2) the comparison of the level of certainty necessary for a finding “beyond a reasonable doubt” with the level of certainty applicable to personal decisionmaking, (3) the suggestion in the fourth paragraph that the jury need only weigh the pros and eons before making a decision, followed by the statement suggesting that whether the decision is right or wrong is of equivalent consequence; (4) the possibility that the third of these paragraphs could be understood as inverting the burden of proof by requiring the jurors to find in the evidence so strong a “conviction” of doubt that they would be able to argue for it to their peers in the jury room. We address each of these in turn, and then also consider their cumulative effect. (1) “Moral Certainty”. Equating the concept of reasonable doubt to “moral certainty” may be, in isolation, reversible error. See Victor, — U.S. at -, 114 S.Ct. at 1250-51; Cage v. Louisiana, 498 U.S. 39, 41, 111 S.Ct. 328, 329, 112 L.Ed.2d 339 (1990) (per curiam); Commonwealth v. Pinckney, 419 Mass. 341, 345-49, 644 N.E.2d 973 (1995). The Supreme Court has discouraged use of this phrase because of its ambiguous meaning, see Victor, — U.S. at -, 114 S.Ct. at 1247-48, and we similarly have expressed concern because “the jury might feel justified in convicting based on a feeling rather than *263 on the facts in the case,” United States v. Drake, 673 F.2d 15, 21 (1st Cir.1982). See also United States v. Indorato, 628 F.2d 711, 721 (1st Cir.1980) (“[W]e have indicated our uneasiness with this phraseology and pointed out that it has been the subject of mixed reviews.”). Indeed, in Cage, the Supreme Court reversed a conviction based on a charge using “moral certainty” language because the only other meaning ascribed to reasonable doubt equated such doubt to “a grave uncertainty” or “an actual substantial doubt.” The Court felt that those terms, in conjunction with the phrase “moral certainty,” suggested a higher degree of doubt than is required for acquittal. In Victor, however, the Court upheld the validity of two separate reasonable doubt instructions that contained “moral certainty” language, observing that that language “cannot be sequestered from its surroundings” and finding that the remainder of the charge lent appropriate content to the otherwise ambiguous words. — U.S. at -, 114 S.Ct. at 1248. As in Victor, the charge here contained far more explanation than was offered to the jury in Cage. The paragraph immediately following the first reference to “moral certainty” distinguishes that level of certitude from mathematical certainty, harking back to the message from the preceding section of the charge. The juxtaposition suggests that the requisite level of confidence was, indeed, substantial, though not proof beyond all doubt. See Pinckney, 419 Mass. at 347, 644 N.E.2d 973. 4 The lengthy charge also offered additional formulations emphasizing the high level of proof necessary for conviction. Twice during the course of the instruction, the court charged that the jury must attain a “settled conviction” of guilt. In Victor, the Supreme Court ruled that the use of a similar phrase, “abiding conviction,” mitigated references to “moral certainty” and “substantial doubt.” See — U.S. at -, 114 S.Ct. at 1247 (“ ‘The word “abiding” here has the signification of settled and fixed, a conviction which may follow a careful examination and comparison of the whole evidence.’ ... As used in this instruction, ... we are satisfied that the reference to moral certainty, in conjunction with the abiding conviction language, ‘impress[ed] upon the factfinder the need to reach a subjective state of near certitude of the guilt of the accused.’ ” (quoting Hopt v. Utah, 120 U.S. 430, 439, 7 S.Ct. 614, 618, 30 L.Ed. 708 (1887) and Jackson v. Virginia, 443 U.S. 307, 315, 99 S.Ct. 2781, 2786, 61 L.Ed.2d 560 (1979)). It also is significant in evaluating the effect of the term “moral certainty” that the jury was told more than once that its decision must be based on the evidence presented. See Victor, — U.S. at -, -, 114 S.Ct. at 1248, 1251. In the third paragraph of this section of the charge, the judge began by noting that “[w]hen you get all through analyzing this evidence, it has to be a doubt nagging your mind, leaving you with an uncertainty of conviction____” Toward the end of the reasonable doubt instruction, the court noted the jurors’ oath to render “a true verdict according to the evidence and the law,” and earlier cautioned against “strain[ing] the evidence to any conclusion not warranted by its fair convincing force.” Thus, as in Victor, the instruction here explicitly told the jurors that their decision had to be based on the evidence in the case, minimizing the possibility that the reference to “moral certainty” would have been viewed as permitting a conviction based “on a feeling rather than on the facts in the case,” Drake, 673 F.2d at 21. See — U.S. at -, 114 S.Ct. at 1248. And, also as in Victor, other instructions reinforced this message. See, e.g., Tr. at 4274 (“We look for a verdict which is dictated by your logic and your common sense and not your heart.”); id. at 4276 (“It is your sworn duty to presume the defendant innocent and to give him the benefit of that presumption all throughout the trial and at every stage of the investigation of the evi *264 denee in the jury room, until it is overcome by proof beyond a reasonable doubt.”); id. at 4281 (“[I]n the last analysis it comes to your most important obligation, and that is, to decide this case on the body of the evidence as I define it.”); id. at 4283 (“[W]hen it comes time to base your verdict, find a foundation in the evidence upon which it must rest.”); id. at 4291 (“The facts must exclude innocence.”). (2) “Vital action in your everyday lives”. Comparing “beyond a reasonable doubt” to the “degree of satisfaction of mind and conscience that jurors should have when they take action in the major affairs of their lives” is an analogy that has drawn criticism for decades. See Drake, 673 F.2d at 20 (noting Supreme Court’s expressed displeasure of the “willing to act” instruction in Holland v. United States, 348 U.S. 121, 75 S.Ct. 127, 99 L.Ed. 150 (1954)). Even when framed in the more accepted format of comparing reasonable doubt to a doubt that would cause a prudent person to hesitate before acting, the instruction is arguably unhelpful. See, e.g., Victor, — U.S. at —, 114 S.Ct. at 1252 (Ginsburg, J., concurring). 5 The instruction here, however, did not include the sort of specific, supposedly comparable, examples that have been viewed as prejudicially misleading to jurors. Compare, e.g., Commonwealth v. Ferreira, 373 Mass. 116, 128-29, 364 N.E.2d 1264, 1272-73 (1977) (reversing because a number of examples of important personal decisions “understated and tended to trivialize the awesome duty of the jury to determine whether the defendant’s guilt was proved beyond a reasonable doubt”) with Rogers v. Carver, 833 F.2d 379, 382-83 (1st Cir.1987) (no reversible error where reference to important decisions was brief and general, with no specific examples) and Bumpus, 635 F.2d at 912-13 (single example of deciding whether to have heart surgery did not trivialize the jurors’ duty or minimize the government’s burden). Because no such examples were used, and because the charge focused on “vital” or “major” personal matters, we think it unlikely that this aspect of the instruction deprived the defendant of the right to be found guilty only upon proof beyond a reasonable doubt. 6 (3) Pros and cons; right and wrong. Read on their own, the opening sentences of the last paragraph in this section unquestionably present an inadequate articulation of the substantial and unique burden of proof born by the prosecution in a criminal case. Read in context, however, the thrust of these passages was to inform the jurors that a “settled conviction of mind” must be reached to find the defendant guilty. Immediately following the reference to right and wrong, the judge stated: But if we are satisfied to a moral certainty when we do an act in our private lives, that it is the right thing to do, we have a settled conviction of mind. That is the degree of proof which the law contemplates when they talk about “proof to a moral certainty.” This explanation makes manifest that the previous comments, though poorly framed, were another restatement of the concept voiced repeatedly by the judge that absolute certainty was unnecessary. The jurors’ decision will not necessarily be error-free: “we may be wrong.” What is crucial, the jurors are told, is whether they can reach a “settled conviction” of guilt. We therefore conclude that the language of this paragraph, though far from ideal, was unlikely to be understood in its entirety in the overly casual way suggested by the opening sentences in isolation. (4) Inversion of burden of proof. The third paragraph of this section of the charge contained two sentences, the first of which told the jurors: you have a reasonable doubt *265 if, when you finish analyzing the evidence, you have a doubt nagging your mind, leaving you with an uncertainty of conviction to that moral certainty which you can stand up and argue in the jury room with principle and integrity and honesty to your fellow jurors. The second sentence was much more direct: “And if you don’t believe in it yourself, you haven’t got a reasonable doubt.” Although the district court viewed the second sentence as a mistake that may have suggested an inversion of the burden of proof, it felt that the preceding sentence “plainly referred to a ‘conviction’ that the defendant was guilty as charged,” not to a “conviction”, i.e., a belief, in a doubt. 866 F.Supp. at 618. It therefore felt that no misimpression was given. Our view is essentially the same. The first sentence clearly refers to the certainty a jury must feel as to conviction. As for the second sentence, the concept of having a belief or a moral certainty in a doubt (which in itself is a state of uncertainty) is, we think, a strange and awkward way of referring to the strength of one’s doubt. The likely effect would have been to confuse, not to encourage an inversion of the burden of proof. In addition, this paragraph was one of twenty in the charge, which began with a statement that “[i]t is the burden of the Commonwealth to establish its case beyond a reasonable doubt,” and which concluded with several paragraphs emphasizing the defendant’s “absolute right to hold the Commonwealth to this strictness of proof.” We therefore find no reasonable likelihood that the jurors entered their deliberations with the false impression that petitioner had the burden of establishing a reasonable doubt. (5) Cumulative effect. As we have discussed, none of the multiple deficiencies in the second portion of the charge was of sufficient magnitude to weaken the conviction. Taken together, however, their effect is more substantial. Reasonable doubt is defined with the imperfect term “moral certainty,” and one alternative explanation of moral certainty is the disfavored formulation concerning personal decisionmaking. Yet another description of reasonable doubt and moral certainty suggests that the jury’s task is simply a matter of weighing the pros and eons to reach a decision that “may be right” or “may be wrong.” A central passage defining reasonable doubt is largely impenetrable, though its language taken literally could be understood to impose a burden of proving doubt on the defendant. And we have expressed our discomfort with the first portion of the charge, which at great length reiterates that the government’s burden should not be overestimated. If these two sections comprised the entire instruction, we might well conclude that reversal would be necessary. Of greatest significance to our contrary conclusion is the fact that, at the conclusion of the portions of the charge we have quoted so far, the judge essentially began anew, telling the jury, “so there will just be no doubt about what reasonable doubt means, I am going to define it in the precise and more scholarly language of our Supreme Judicial Court.” In the next eight paragraphs, he presents the then-acceptable charge on reasonable doubt from Commonwealth v. Madeiros, 255 Mass. 304, 307-08, 151 N.E. 297 (1926), see Pinckney, 419 Mass. at 348, 644 N.E.2d 973, together with language emphasizing the importance of the reasonable doubt standard to our system of jurisprudence. 7 “Proof beyond a reasonable doubt does not mean proof beyond all doubt, nor beyond a whimsical or a fanciful doubt, nor proof beyond the possibility of innocence. “It is rarely, if ever possible, to find a case so clear that there cannot be a possibility of innocence. If an unreasonable doubt or a mere possibility of innocence were sufficient to prevent a conviction, *266 practically every criminal would be set free to prey upon the community. Such a rule would be wholly impracticable and would break down the forces of law and order, and make the lawless supreme. “A reasonable doubt does not mean such doubt as may exist in the mind of a man who is earnestly seeking for doubts or for an excuse to acquit a defendant; but it means such doubt as remains in the mind of a reasonable man who is earnestly seeking the truth. “A fact is proved beyond a reasonable doubt when it is proved to a moral certainty, as distinguished from an absolute or mathematical certainty; when it is proved to a degree of certainty that satisfies the judgment and consciences of the jury as reasonable men, and leaves in their minds, as reasonable men, a clear and settled conviction of guilt. But if when all is said and done there remains in the minds of the jury any reasonable doubt of the existence of any fact which is essential to the guilt of the defendant on the particular charge, the defendant must have the benefit of it and cannot be found guilty upon that charge.” And if you have a reasonable doubt, your verdict ought to be not guilty. I want to say a few more words about reasonable doubt. A standard which produces great satisfaction to me, and I think should to you, the defendant has an absolute right to hold the Commonwealth to this strictness of proof. No consideration of public safety, nor righteous indignation at atrocious crime which shocks the community, nor zeal for the suppression of crime can give to the Court and jury the discretion, or the right to relax this standard of proof; nor to strain the evidence to any conclusion not warranted by its fair convincing force. This is a government of laws and not of men. If the guilty go unpunished today because jurors observed their oath and rendered a true verdict according to the evidence and the law, then the community and every citizen in it is still safer, because the law has prevailed. The Court cannot state this principle too strongly as a principle to be observed, to guard the rights of a defendant. And I would be remiss in my duty if I did not with equal force remind you that the community is not safe if only the rights of those charged with crime are safeguarded and protected and the rights of the Commonwealth to have a verdict if guilt is proved beyond a reasonable doubt is just as absolute and just as sacred as the right of a defendant to an acquit[t]al if the proof does not meet that test. To the extent that specific portions of the instruction up to this point had been less than clear, the jury explicitly was told that this restatement was equivalent and complete. Although the judge repeated in this part of the instruction the “moral certainty” phraseology, he contrasted a moral certainty only with an absolute or mathematical certainty. See Pinckney, 419 Mass. at 347, 644 N.E.2d 973 (finding that the identical language “properly impressed upon the jury the need to reach a subjective state of near certitude of the guilt of the accused”). See also swpra at 262-63. Moreover, the judge again emphasized that the proof must leave “reasonable men” with “a clear and settled conviction of guilt,” and, failing that, the defendant must be found innocent. With the language discouraging a verdict for the prosecution unless based on the law and the evidence, this alternative charge was entirely correct. In our view, even the cumulative effect of the earlier imperfections was offset by this lengthy and independent charge, which the judge praised to the jury as “precise and more scholarly.” In sum, this instruction possessed a number of flaws, as did the instructions reviewed in Victor. As we look at some of the less defensible language, we find it difficult to say that a juror could not have been led astray. But as the Court reminded us in Victor, — U.S. at -, 114 S.Ct. at 1248, the standard is not “could have” but rather: is there a reasonable likelihood that the jury understood the instruction as a whole to permit conviction based on a level of proof below that required by the Due Process Clause? Our review convinces us that there was no such likelihood in this case. Accordingly, it *267 provides no basis for disturbing Gilday’s conviction. III. Brady/Giglio Claims Gilday contends that the government’s failure to disclose cooperation agreements with two accomplices who testified as prosecution witnesses, and the failure to correct their false testimony that no deals were made, violated his due process rights as established in Brady v. Maryland, 373 U.S. 83, 87, 83 S.Ct. 1194, 1196-97, 10 L.Ed.2d 215 (1963) and Giglio v. United States, 405 U.S. 150, 154-55, 92 S.Ct. 763, 766, 31 L.Ed.2d 104 (1972). 8 He additionally raises separate Brady claims based on the government’s failure to disclose exculpatory statements by an eyewitness to the crime who did not testify at trial and by two trial witnesses. We have examined these claims with care, and have concluded that none of the asserted nondisclosures nor all of them cumulatively constitute reversible error. We begin our discussion by noting the relevant standards. A Brady error occurs when the prosecution suppresses “material” evidence that is favorable to the accused. See Kyles v. Whitley, — U.S. ——, —, 115 S.Ct. 1555, 1565, 131 L.Ed.2d 490 (1995). In most circumstances, exculpatory evidence is material 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,’ ” id. (quoting United States v. Bagley, 473 U.S. 667, 682, 685, 105 S.Ct. 3375, 3383, 3385, 87 L.Ed.2d 481 (1985)). 9 We refer to this as the Bagley standard. A standard of materiality more favorable to the defendant applies, however, when previously undisclosed evidence reveals that the prosecutor knowingly used perjured testimony or, “equivalently,” knowingly failed to disclose that testimony used to convict the defendant was false. Bagley, 473 U.S. at 678-80, 105 S.Ct. at 3381-82. In such situations, “ ‘a conviction ... is fundamentally unfair, and must be set aside if there is any reasonable likelihood that the false testimony could have affected the judgment of the jury,’ ” Kyles, — U.S. at — n. 7, 115 S.Ct. at 1565 n. 7 (quoting United States v. Agurs, 427 U.S. 97, 103, 96 S.Ct. 2392, 2397, 49 L.Ed.2d 342 (1976)) (emphasis added). 10 We shall label this test the Agurs standard. Although the tests for materiality suggest a harmless error-like inquiry, it is important to note that these standards must be applied to determine the threshold question: has constitutional error occurred? Only then does the issue of harmlessness arise. And, as the Supreme Court’s recent decision in Kyles makes clear, see — U.S. at —, 115 S.Ct. at 1567, the approach to harmless error in the Brady/Giglio context has evolved as the Chapman formulation of “harmless beyond a reasonable doubt” has yielded in habeas cases to the softer Brecht test of whether the error “ ‘had substantial and injurious effect or influence in determining the jury’s verdict,’ ” Brecht v. Abrahamson, — U.S. —, —, 113 S.Ct. 1710, 1722, 123 L.Ed.2d 353 (1993) (quoting Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 1253, 90 L.Ed. 1557 (1946)). In Kyles, the Court observed that harmless error analysis is inapplicable to a Brady/Giglio claim arising in a habeas case outside the perjury-related context. — U.S. at — & n. 7, 115 S.Ct. at 1565 & n. 7 (noting that “our decision today does not address any *268 claim under the first Agurs category [i.e., perjury-related]”). The reason is compelling: the Bagley materiality standard necessarily requires a court to find an impact on the jury verdict sufficiently substantial to satisfy the Brecht harmless error test. Thus, in the non-perjury setting, all that is required or appropriate is the one-step Bagley inquiry into reasonable probability. But a prosecutor’s knowing use of false testimony presents a different analytical situation. As Bagley makes clear, a petitioner is given the benefit of a friendly standard (hostile to the prosecution) to establish materiality: whether a reasonable jury could have been affected. 473 U.S. at 678-80, 105 S.Ct. at 3381-82. This is, in essence, the old Chapman inquiry. Id. at 679-80 & n. 9, 105 S.Ct. at 3382 & n. 9. Applying this standard in most cases involving perjury or its equivalent will likely result in a finding of constitutional error. Scaling that lower materiality hurdle, however, still will leave the petitioner facing the Brecht harmless error inquiry into whether the perjured testimony in fact had a substantial and injurious effect or influence on the jury’s verdict. 11 In other words, where the Agurs standard applies, it is quite possible to find a constitutional violation, but to conclude that it was harmless. When faced with such a claim, therefore, our inquiry is necessarily two-pronged: was there a failure to disclose material exculpatory evidence, and, if yes, was such failure harmless? Having laid out this framework, we now turn to petitioner’s claims. A. Prosecutorial agreements with Fleischer and Valeri Petitioner argues that the government deliberately relied on the false testimony of two witnesses, Fleischer and Valeri, who denied that any deals had been made with the prosecution for their cooperation. The district court, like the Commonwealth courts before it, concluded that no error occurred with respect to Valeri because his acknowledgement at trial of “a generalized expectation of leniency” (i.e., that his cooperative trial testimony would be brought to the court’s attention) served to disclose his possible motivation to testify favorably for the government. We agree with this determination essentially for the reasons expressed by the district court, and do not address it further. See 866 F.Supp. at 634-36. The Supreme Judicial Court did conclude, however, that the government improperly failed to disclose a deal made with the attorney for the other witness, Fleischer. Called in rebuttal, Fleischer testified most crucially that, in a discussion shortly after the robbery, Saxe and Power accused Gilday of being “trigger-happy” and that Gilday said, “What did you want me to do, the cop was right there, he was only thirty seconds behind you.” In cross-examination, Fleischer specifically denied that any deals had been made for his testimony, and testified further that his only promise from the Commonwealth was that high bail would be requested but not demanded. In fact, as found by a Superior Court judge following a hearing on petitioner’s motion for new trial, the prosecutor had told Fleischer’s attorney that, in exchange for Fleischer’s testimony, he would attempt to reach a disposition of the charges against Fleischer that would leave him with no criminal record. 12 Commonwealth v. Gilday (Gilday II), 382 Mass. 166, 175, 415 N.E.2d 797, 802 (1980). The motion judge also found that, as the prosecutor and Fleischer’s attorney had agreed, Fleischer was not told of this specific arrangement; he knew only that it would be “in his best interest to testify in the case.” Id. The Supreme Judicial Court noted that neither the lack of a formal agreement with Fleischer nor Fleischer’s lack of knowledge of the specifics of the understanding relieved *269 the prosecutor of his obligation to disclose material facts concerning Fleischer’s credibility and possible bias. Indeed, the court noted that to hold otherwise would be “in effect [to] approve the evasion of the Giglio rule by means of artful device.” 382 Mass. at 177, 415 N.E.2d at 803. That court, however, agreed with the motion judge that the suppression of this information had no effect on the jury and therefore did not require a new trial. 382 Mass. at 177-78, 415 N.E.2d at 803-04. The district court reached the same conclusion in rejecting the habeas petition. 866 F.Supp. at 633. 13 Our review of the determination that the prosecutor had a duty to disclose the Fleischer arrangement is de novo. See Ouimette, 942 F.2d at 4. We apply the Agurs standard of materiality, more favorable to the petitioner, because of the prosecutor’s deliberate strategy to misrepresent Fleischer’s credibility and the knowing acquiescence in Fleischer’s false testimony. See id. at 11; see also supra at 267. As we have explained, the relevant inquiry is whether there is any reasonable likelihood that the false testimony could have affected the judgment of the jury. To put the question the other way around: can we say that no reasonable jury could have been affected by the undisclosed information? We think the answer is fairly obvious. The information withheld by the prosecutor would have provided the basis for powerful impeachment of Fleischer’s testimony. Not only did Fleischer deny that any deals had been struck on his behalf, but he also claimed that he was testifying only because a man had been killed and he wanted to “see justice done.” The fact that his lawyer and the prosecutor had come to an understanding would have markedly strengthened the defense’s claim that Fleischer was highly motivated to implicate Gilday to protect himself. First, it would have permitted the jury reasonably to infer that, even if the “wink and nod” deal had not been explicitly communicated to Fleischer, he must have been given some indication that testimony helpful to the government would be helpful to his own cause. Cf. Bagley, 473 U.S. at 683, 105 S.Ct. at 3384 (making reward contingent upon outcome “served only to strengthen any incentive to testify falsely in order to secure a conviction”). In addition, evidence of the deal would have reinforced the testimony of defense witness Bond, another accomplice, who implicated Fleischer as the gunman. The stakes for Fleischer were substantial indeed if his testimony blaming someone else could secure his release entirely from criminal responsibility for a murder he had committed; his motivation to lie could not have been greater. Disclosure of the deal in all likelihood would have reduced substantially, or even destroyed, Fleischer’s credibility. Because the direct accusation of an accomplice is of more than minimal consequence in a case where the defense is that someone else was responsible for the charged crime, we think it at least reasonably likely that the suppression of this evidence could have affected the jurors’ judgment. Presumably, the government agrees with this assessment; for what other reason would the prosecutor have gone to such lengths to keep the information from them? Recognition of error does not end our task, however. Although we have determined that the jury might have been affected by knowledge of Fleischer’s deal, and thus that the prosecution’s suppression of the evidence violated its constitutional obligation under Brady and Giglio, we also must consider — to restate the Brecht standard — whether the error was of such magnitude that it actually casts doubt on the integrity of the verdict. This is the difference between a possibility and a probability. See O’Neal v. McAninch, — U.S. -, -, 115 S.Ct. 992, 994, 130 L.Ed.2d 947 (1995) (to find harmlessness, reviewing court must conclude that error more likely than not had no effect on the verdict). Our review of the evidence indicates that, even if the jury had assigned no weight to Fleischer’s testimony, the sub *270 stance of the case against Gilday would have remained the same. The other . evidence, moreover, was considerable. Cf. Giglio, 405 U.S. at 154-55, 92 S.Ct. at 766 (government’s case depended “almost entirely” on witness whose deal with prosecution was not disclosed). Indeed, Fleischer was a rebuttal witness, and as such simply repeated the earlier testimony of another witness, Valeri, that Gilday had admitted to being the shooter. Valeri had reported that Gilday was at the scene of the robbery in the car from which the shots were fired, that Gilday possessed the murder weapon after the crime, and that Gilday said that he had waited at the scene of the robbery until the police officers arrived because “he had always wanted to shoot a police officer.” Three eyewitnesses testified, all disinterested outsiders who were in close proximity to the shooting. The strongest, Becker, made an in-court identification of Gilday, who was sitting unobtrusively with spectators. He also had chosen Gilday’s picture from a spread of photographs shown to him two months after the shooting. He further recalled at trial, after having stood up to make his identification, that the gunman, like himself (and like Gilday), had a little bald spot on the top of his head. Cross-examination elicited that the only description Becker had given police at the time of the shooting was that the gunman was a white male and that he probably had seen Gilday’s picture in the media; while acknowledging the possibility that this influenced him, he insisted that his identifications were based solely on his observations at the time of the crime. A second witness, Goddard, described the gunman as a white male in his late thirties, clean-shaven, with a receding hairline, hair combed straight back, wearing an olive jacket — all consistent with Gilday’s appearance. True, he did not pick out Gilday’s picture from spreads shown him on the day of the shooting, or two weeks later. On the first occasion, he saw “a couple of pictures that resembled the man that I saw” but did not pick them out because he was not sure. On the second occasion, he had seen Gilday’s picture in the papers and recognized it in the spread. The third witness, Gaudette, described the gunman as of medium build, similar to himself, with weight around 185 pounds, height five feet six or seven inches, in his thirties, with dark hair. He picked Gilday’s picture out of a photospread two months after the shooting. Then, in the courtroom, he failed to identify Gilday, who was not wearing glasses at the start of Gaudette’s perusal but put them on as Gaudette continued his scrutiny. On the whole, we consider this eyewitness evidence, while not without weaknesses, impressive. Further, the evidence was overwhelming that Gilday was involved in the crime. In his own testimony, he acknowledged that he had bought the semiautomatic weapon and one of the cars that had been used in the robbery and murder, that he had stolen a license plate and affixed it to another car used by the robbers, and that he took some of the holdup money from an apartment where Bond, Saxe, Power and Fleischer were gathered after the crime. In addition, an acquaintance of the group, McGrory, testified to a conversation with Gilday after the robbery about McGrory’s having figured out who was responsible for the crime. McGrory stated that, during the exchange, Gilday first claimed that he had not heard that a police officer had been critically wounded, but later said “I did it” and warned that even if Gilday were imprisoned on death row, he would take care of McGrory if he said anything. While neither the activities to which Gilday confessed nor McGrory’s testimony require a finding that Gilday was the gunman 14 — or even at the scene of the robbery — the other evidence provided a strong link between his confessed complicity in the crime and the alleged role of gunman. Fleischer’s testimony on rebuttal corroborated the government’s case, but it shed no new or different light on it. *271 Moreover, some impeachment of Fleischer did take place. A former FBI agent, George Bernard Kennedy, testified in surrebuttal that Fleischer had told him in the spring of 1971 that Gilday had not shot the police officer. In addition, as we have noted, Fleischer was implicated by defense witness Bond as the gunman; that testimony made transparent a motive to accuse someone else. Thus, while the challenge to Fleischer’s credibility would have been measurably more potent with the withheld information, he was already a sullied witness. In sum, we are persuaded that suppression of the Fleischer agreement did not have a substantial and injurious effect or influence in determining the jury’s verdict. The nature of the other evidence makes it unlikely that even a stinging impeachment of Fleischer based on the undisclosed deal would have had the requisite impact on the jury’s deliberations. 15 The error was therefore harmless. 16 B. Eyewitness and other statements. Petitioner also claims error in the government’s failure to disclose exculpatory evidence from three witnesses: (1) a statement by Michael Finn shortly after the crime that described the gunman in terms inconsistent with Gilday’s appearance; (2) an FBI report that Bernard Becker, who identified Gilday as the gunman at trial, stated three weeks after the crime that he could not provide a description of the gunman; (3) a report that Fleischer initially denied knowing Gilday when asked by the FBI. To determine the materiality of these pieces of undisclosed evidence, we apply the Bagley test: whether there is “ ‘a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different,’ ” Kyles, — U.S. at —, 115 S.Ct. at 1565 (quoting Bagley, 473 U.S. at 682, 105 S.Ct. at 3383). Only two merit more than passing discussion. 17 While recognizing that Finn’s first description of the gunman should have been viewed by the prosecutor at the outset of the trial as significant exculpatory evidence subject to disclosure, the vantage of hindsight leads us to conclude that the likely impact of suppressing it turned out to be slight. At the hearing on petitioner’s second motion for new trial, which was premised in part on this asserted Brady violation, Finn, who was not called as a witness at trial, gave a different description of the shooter—one consistent with Gilday’s appearance. The motion judge found that Finn, who had a reputation for mental instability and alcoholism, would have been an unreliable witness. Gilday I, 367 Mass, at 487-89, 327 N.E.2d at 859-60. The district court credited this finding, 866 F.Supp. at 636, and we are likewise inclined to do so. Based on the record before us, it seems most likely that, had Finn testified at trial, his impact on the jury would have been at best equivocal. We certainly cannot say that there is a reasonable probability that his testimony would have materially changed the face of the trial. As for Becker, while we appreciate that statements made in close temporal proximity to the crime are significant in evaluating an eyewitness’s reliability, see Kyles, — U.S. at —, 115 S.Ct. at 1569 (citing Manson v. Brathwaite, 432 U.S. 98, 114, 97 S.Ct. 2243, 2253, 53 L.Ed.2d 140 (1977)), we think it only of modest importance that his identification of petitioner was preceded by an earlier inability to provide a description. The fact that someone cannot articulate a description of an individual does not necessarily *272 undermine the accuracy of a later identification; the challenge would be substantially more potent had Becker given an earlier description inconsistent with his trial testimony. Moreover, Becker’s testimony at trial essentially reflected the fact that he was unable to give a description at the time of crime. He stated that he told the police only that the shooter was “[a] white male,” acknowledging that he provided “[n]o other distinguishing characteristics or descriptions.” In these circumstances, we think the withheld FBI report would have had a minimal effect upon the jury. C. Cumulative Impact. In its recent decision in Kyles v. Whitley, the Supreme Court stressed the importance of considering the cumulative effect of all suppressed evidence in determining whether a Brady violation has occurred. — U.S. at -, 115 S.Ct. at 1567. The Court concluded that, had the prosecution disclosed to competent counsel the substantial amount of evidence at issue there, a different result would have been reasonably probable; 18 the jury would have been considerably more likely to have accepted the defendant’s theory that he was framed by a police informant who was actually the murderer. See id. at -, 115 S.Ct. at 1562-64. 19 Not only would disclosure have severely discredited two of the four eyewitnesses who were “ ‘the essence of the State’s ease,’ ” id. at -, 115 S.Ct. at 1559 (quoting district court), but it also would have “entitled a jury to find that ... the most damning physical evidence was subject to suspicion, that the investigation that produced it was insufficiently probing, and that the principal police witness was insufficiently informed or candid.” Id. at -, 115 S.Ct. at 1574. The circumstances here are markedly different. While the various pieces of suppressed evidence in Kyles fit together factually to make the defense theory of the case more likely, the evidence here taken cumulatively sheds no new light on the crime or petitioner’s involvement in it. The suppressed material went primarily to the credibility of witnesses, one of whom (Becker) acknowledged at trial information equivalent to the undisclosed evidence. The only significantly potent undisclosed material was the Fleischer agreement; 20 we are sufficiently persuaded that none of the other evidence adds to its effect in such a way as to have influenced the jury’s judgment. Here, unlike in Kyles, the whole of the challenge to the prosecution’s case was no greater than the sum of its individual parts. Accordingly, we find no remediable Brady violation. IV. Sandstrom Claims Petitioner argues that the trial judge’s charge to the jury included five mandatory presumptions of intent that violated his due process rights as established in Sandstrom v. Montana, 442 U.S. 510, 520-24, 99 S.Ct. 2450, 2457-59, 61 L.Ed.2d 39 (1979). 21 In a *273 related argument, he challenges a portion of the instruction foreclosing the jury from considering an intoxication defense, arguing that it unconstitutionally relieved the Commonwealth of its burden of proof on the issue of intent as recognized by the Supreme Judicial Court in a series of eases beginning with Commonwealth v. Henson, 394 Mass. 584, 592-93, 476 N.E.2d 947, 953-54 (1985). 22 The SJC reviewed the instructions on intent only as they related to the issue of intoxication. The court ruled that the intoxication portion of the charge correctly reflected the law at the time of petitioner’s trial, and that he was not entitled to retroactive application of a change in the law that was announced thirteen years later. See Commonwealth v. Gilday (Gilday III), 409 Mass. 45, 47, 564 N.E.2d 577, 579 (1991). The court found that no other assertion of instructional error, including the more general Sandstrom claims, had been raised in the relevant (fourth) motion for new trial. It therefore held that such additional claims were waived. Id., 409 Mass. at 46 & n. 3, 564 N.E.2d at 578 & n. 3. The district court did not explicitly address the intoxication issue. Petitioner raises it on appeal in limited fashion, recognizing that we previously have declined to disturb the SJC’s determination that Henson’s protection of the intoxication defense is not retroactive. See Robinson v. Ponte, 933 F.2d 101, 103-05 (1st Cir.1991). We decline to revisit that precedent here, and the claim is therefore unavailing. 23 As for the general Sandstrom claims, the district court gave two reasons for rejecting them. First, it concluded that petitioner was not entitled to protection from the Sandstrom rule because that case was decided after his conviction became final, and, under Teague v. Lane, 489 U.S. 288,109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), the principle established there was not retroactive. Second, the court held that the SJC’s refusal to consider the “non-intoxication claims” based on petitioner’s procedural default constituted an independent state law ground for rejecting those claims, thus barring habeas review unless the petitioner can show “cause for”, and “prejudice from” his noncompliance with the Commonwealth’s procedures. See Wainwright v. Sykes, 433 U.S. 72, 86-87, 97 S.Ct. 2497, 2506, 53 L.Ed.2d 594 (1977); Ortiz v. Dubois, 19 F.3d 708, 714 (1st Cir.1994). 24 We need not delve into the retroactivity issue because we agree with the district court’s judgment that the non-intoxication Sandstrom claims are proeedurally barred. Petitioner’s fourth motion for new trial and his memorandum in support of the motion focused entirely on the effect of the intoxication instruction on the jury’s determination of intent. Although the memorandum cited to Sandstrom and related precedent, i.e., In re Winship, 397 U.S. 358, 90 S.Ct. 1068, 25 L.Ed.2d 368 (1970) and Mullaney v. Wilbur, 421 U.S. 684, 95 S.Ct. 1881, 44 L.Ed.2d 508 (1975), it is apparent to us that those cases were invoked in support of the intoxication argument and not as a basis for a general challenge to the presumptions on intent contained in the instruction. In these *274 circumstances, we cannot contradict the SJC’s determination that, as a matter of Massachusetts law, the motion did not preserve the Sandstrom claims for appeal. See Ortiz, 19 F.3d at 713 n. 5 (“the law of Massachusetts is what the SJC says it is”). Cf., e.g., Williams v. Lane, 826 F.2d 654, 660 (7th Cir.1987) (state court determination of waiver does not preclude federal habeas review where record shows that petitioner fully complied with state’s articulated procedural rules). Accordingly, we do not consider them. 25 V. Pretrial Publicity Petitioner contends that he was denied his right to a fair trial because of extensive pretrial publicity, specifically claiming that the trial judge erred in denying his motion for an additional continuance of the trial date and for a change of venue. 26 Essentially for the reasons expressed by the Supreme Judicial Court and the district court, we find no reversible error in the trial judge’s handling of the case in this respect. See Gilday I, 367 Mass. at 491-93, 327 N.E.2d at 861-62; Gilday v. Callahan, 866 F.Supp. at 623-24. VI. Conclusion We have examined with care each of petitioner’s claims of constitutional error. Having found that the only meritorious claim— the Brady violation in suppressing the Fleischer agreement — was harmless, we affirm the judgment of the district court denying petitioner’s writ of habeas corpus. Affirmed. Footnotes [1] The other defendants charged in the crime were Stanley R. Bond, Robert J. Valeri, Susan E. Saxe and Katherine A. Power. Michael Fleischer was charged as an accessory after the fact. Bond, who testified as a defense witness at Gil-day’s trial, died in prison. Valeri testified as a Commonwealth witness against Gilday, pled guilty eight months later to manslaughter, and is now free. Fleischer also testified as a Commonwealth witness, and his indictments ultimately were dismissed. Saxe was a fugitive for several years. After her first trial ended with a hung jury, she pled guilty to manslaughter and is now free. Power surrendered to authorities in 1993, and is now serving a prison sentence. [2] The Bumpus trial and Gilday's both occurred early in the 1970s, and the same judge presided over them. [3] All nineteen substantive paragraphs are contained in one of the three segments of the instruction set off in blocks from the text of this opinion. The twentieth, which simply introduces the final portion of the charge, is reproduced in text on pages 15 and 16. [4] Although that same distinction was drawn in the reasonable doubt instruction in Cage, the charge there did not elaborate any further and, to the extent it did provide additional explanation, could not overcome the "grave uncertainty” and "actual substantial doubt” language that the Court found unacceptable. [5] The Supreme Court repeatedly has approved the “hesitate to act” formulation, however, and the majority in Victor relied in part on the trial court's use of it as an alternative definition of reasonable doubt to support its conclusion that the instruction there was adequate. The Court noted that it “gives a common-sense benchmark for just how substantial such a doubt must be.” — U.S. at —, 114 S.Ct. at 1250. [6] The judge's third reference to everyday decisionmaking was not qualified with an adjective such qs "vital” or "major,” but we think the need to equate the criminal trial with a matter of grave importance was by that time clear. [7] The Supreme Judicial Court has since criticized the Madeiros language, in the second paragraph quoted here, as warning the jury against holding the prosecution to too high a standard of proof. See Commonwealth v. Pinckney, 419 Mass. 341, 348, 644 N.E.2d 973 (1995) (citing cases). In this case, the judge remedied that deficiency later in the instruction by warning the jury against relaxing the reasonable doubt standard in response to concerns about public safety or shocking crimes. [8] Brady established a prosecutor's obligation to turn over exculpatory material. In Giglio, the Supreme Court held that the obligation includes evidence that would impeach the credibility of government witnesses. [9] This standard applies when the government failed to respond to a specific defense request for exculpatory evidence, and when the government failed to volunteer exculpatory evidence never requested, or requested only in a general way. Kyles v. Whitley, - U.S. -, -, 115 S.Ct. 1555, 1565, 131 L.Ed.2d 490 (1995). [10] Kyles explicitly refers only to the knowing use of perjured testimony, but we think it implicit that the Court also contemplated application of this test to those "equivalent" circumstances noted in Bagley. We have applied the Agurs standard in a non-perjuiy setting, when a prosecutor intentionally withheld materials relating to a witness’s prior criminal record and to the deals he made with the state. See Ouimette v. Moran, 942 F.2d 1, 10-11 (1st Cir.1991). [11] It is the government's burden, of course, to demonstrate that the error is harmless. O’Neal v. McAninch, - U.S. -, -, 115 S.Ct. 992, 995-98, 130 L.Ed.2d 947 (1995). [12] Petitioner states in his brief, although without citation, that the nine indictments against Fleischer, including accessory after the fact to first degree murder, in fact were dropped after he testified against Gilday and Saxe in their separate trials. The Commonwealth does not dispute this representation. [13] Neither the Supreme Judicial Court nor the district court found it necessaiy to state precisely whether the applicable Brady standard had been met and a constitutional violation thus established because each found the asserted error harmless in any event. [14] Gilday maintains that none of this testimony explicitly refers to his being the gunman, but simply indicates participation generally in the group's pursuits. [15] Contrary to petitioner's suggestion, the references to Fleischer in the prosecutor’s closing argument did not particularly highlight his testimony. Indeed, Fleischer was significant in the prosecution's summary of the case in part because defense witness Bond indicated that he was the gunman. [16] Petitioner urges that this is the "unusual case" identified in Brecht as perhaps warranting habeas relief even if the asserted error did not substantially influence the jury’s verdict. See-U.S. at — n. 9, 113 S.Ct. at 1722 n. 9. We do not find that "the integrity of the proceeding” was so infected here as to justify such an extraordinary remedy. [17] The failure to disclose Fleischer's false statement is measurably less significant to his credibility than the cooperation agreement, and our earlier discussion concerning Fleischer’s testimony effectively resolves this issue as well. [18] Indeed, a mistrial was declared in the defendant's first trial because the jury became deadlocked on the issue of guilt. - U.S. at -, 115 S.Ct. at 1562. [19] The undisclosed evidence in Kyles consisted of six contemporaneous eyewitness statements with significant inconsistencies; records of the informant's initial call to the police stating that he had bought a car like the victim’s from the defendant on the day of the murder; a tape recording of another conversation between the informant and police officers further incriminating the defendant; a signed statement from the informant repeating matters from the tape-recorded discussion, although with inconsistencies; a computer print-out of license numbers of cars parked on the night of the crime at the market where the murder occurred, which did not list the defendant’s car license number; an internal memorandum directing seizure of the defendant’s trash after the informant had suggested the victim's purse might be found there, and evidence linking the informant to other crimes at the market and to an unrelated murder. [20] Although we found error in the prosecution's suppression of the Fleischer agreement, our conclusion that the error was harmless effectively is equivalent to our finding that the other evidence withheld did not satisfy the Bagley materiality standard. We therefore reconsider the agreement in assessing the cumulative effect of the government’s non-disclosures. [21] Sandstrom held that a jury instruction containing a presumption that has the effect of relieving the prosecution of the burden of proof on an element of a charged crime violates the Due Process Clause. 442 U.S. at 520-24, 99 S.Ct. at 2457-59. [22] The court stated in Henson that "where proof of a crime requires proof of a specific criminal intent and there is evidence tending to show that the defendant was under the influence of alcohol or some other drug at the time of the crime, the judge should instruct the jury, if requested, that they may consider evidence of the defendant's intoxication at the time of the crime in deciding whether the Commonwealth has proved that specific intent beyond a reasonable doubt.” 394 Mass. at 593, 476 N.E.2d at 954. [23] To the extent that petitioner seeks to raise the federal constitutional claim directly on appeal, we decline to address the issue. His brief contains only two paragraphs on the intoxication instruction, primarily directed to the Supreme Judicial Court's caselaw and our decision in Robinson v. Ponte. His reference to arguments raised in the district court, without elaboration, is insufficient to warrant our review. See Cray Communications v. Novatel Computer Systems, 33 F.3d 390, 396 n. 6 (4th Cir.1994) (adopting by reference memoranda filed in the district court is a practice "that has been consistently and roundly condemned by the Courts of Appeals”); United States v. Bales, 813 F.2d 1289, 1297 (4th Cir.1987) (noting that "other courts have stated that arguments incorporated by reference need not be considered on appeal”). [24] Petitioner has not argued that there was "cause for” or "prejudice from” his failure to comply with the procedural rule, relying instead on the assertion that no default occurred. [25] We suspect, however, that even if considered on the merits, the Sandstrom claims would be deemed harmless error. Petitioner’s defense was not that he lacked the requisite mens rea to be found guilty on the crimes charged, but that he was not the gunman who shot Officer Schroeder. See Bembury v. Butler, 968 F.2d 1399, 1402-1404 (1st Cir.1992) (instruction creating mandatory presumption of intent was harmless where question of intent never raised: "[Defendant] merely presented an alibi, claiming he was not the culprit.”) [26] The trial originally was scheduled to start in April 1971, about six months after the crime, but the court granted a continuance and it did not begin until February 1972.