The defendant appeals from his conviction of breaking and entering a dwelling house in the nighttime with intent to commit larceny. He took no exceptions to the judge’s charge to the jury, but he argues to us that the instructions on reasonable doubt and on presumptions require reversal to prevent a miscarriage of justice. We affirm the conviction.

*597The defendant was indicted in June, 1976, for a burglary on April 1,1976, and was tried before a jury with two codefendants, Edward and George Mallett, in September, 1976. He was sentenced to a prison term of four to six years, to be served from and after a sentence he was then serving. The defendant’s petition for leave to file a late bill of exceptions was allowed by a single justice of this court in September, 1978, and the trial judge later allowed his motion to make the trial subject to G. L. c. 278, §§ 33A-33H. After the defendant’s appeal was entered in the Appeals Court, we allowed his application for direct appellate review.

We summarize the evidence on behalf of the Commonwealth. About 10 p.m. on April 1, 1976, a Worcester police officer observed the defendant and Edward Mallett loading furniture in a van in front of a six-family building. George Mallett stayed by the van. Police officers followed the defendant and Edward Mallett to a third-floor apartment. The front door of the apartment had been forced open, the outside lock was hanging on the door, part of the lock was on the floor, and the door had been broken. The bedroom had been ransacked. The police arrested the three men. Karle Lamson, the manager of the building and resident of the apartment, testified that he was on Cape Cod at the time. A few days before an intruder had broken the door of the apartment, and the defendant put on a new door and transferred the lock. The defendant was “working as kind of an assistant manager with me in the building.” According to Lamson’s testimony, he asked Mark Rossiter to check his apartment daily while he was away “to see if anything was wrong,” but he did not give anyone permission to enter his apartment or to remove any furniture during his absence.

Rossiter testified for the defense that he had been asked by Lamson to drop by and check the apartment, that he was aware of a similar understanding between Lamson and the defendant, who lived on the floor below Lamson, and that he had discovered a second forcible entry into the Lamson apartment, had made a list of missing items, and had fixed the door jamb so the door would close but not lock. Rossiter *598said he talked to the defendant about the second break and they agreed “to sort of stay out of it” until Lamson came back. The defendant testified to the two breaks and to the conversation with Rossiter, adding that Lamson’s girl friend had said Lamson wanted the defendant to watch the apartment while Lamson was away. After he had talked to Rossiter, according to the defendant, he discussed the situation with Edward Mallett. They decided to enlist the help of Edward’s brother George and his van and to move Lamson’s remaining furniture to Mallett’s mother’s house for safekeeping. To impeach the defendant’s credibility, evidence was admitted of his conviction in 1969 of five crimes, including armed robbery and burglary.

Edward Mallett did not testify. George Mallett testified that his brother Edward telephoned him about 8:30 p.m. on the evening in question, said he had bought some used furniture for their mother, and asked for a ride to pick it up. George later picked up his brother, drove him to the apartment house, and stayed by the van while Edward and the defendant carried out furniture.

The jury found the defendant and Edward Mallett guilty as charged, but acquitted George Mallett. The defendant made no objections and took no exception to any portion of the judge’s charge, but argues to us that the cumulative errors in the judge’s charge on proof beyond a reasonable doubt and on proof of specific intent to commit larceny resulted in a miscarriage of justice, citing Commonwealth v. Freeman, 352 Mass. 556, 563-564 (1967).

1. Reasonable doubt. The defendant claims error in the judge’s instruction to the jury on reasonable doubt in three respects: (1) the use of a decision whether to undergo heart surgery as an illustration; (2) the charge that a reasonable doubt was a doubt for which a juror could give a “good reason”; (3) warning of the consequences of holding the Commonwealth to too strict a standard, without counterbalancing warnings about the danger of convicting an innocent person.

Each of these claims has a foundation in our decisions. (1) In Commonwealth v. Bumpus, 362 Mass. 672, 682 *599(1972) , judgment vacated on other grounds, 411 U.S. 945 (1973) , aff’d on rehearing, 365 Mass. 66 (1974), reviewed on petition for habeas corpus, sub nom., Bumpus v. Gunter, 452 F. Supp. 1060 (D. Mass. 1978), we identified the “inherent difficulty” in using examples such as a decision to undergo major heart surgery: “while they may assist in explaining the seriousness of the decision before the jury, they may not be illustrative of the degree of certainty required.” See Commonwealth v. Ferguson, 365 Mass. 1, 12 (1974); Commonwealth v. Coleman 366 Mass. 705, 712 (1975). In Commonwealth v. Ferreira, 373 Mass. 116, 128-130 (1977), where “the evidence of the defendant’s guilt was not overwhelming,” we reversed a conviction of murder in the first degree pursuant to G. L. c. 278, § 33E, in part because the judge’s instructions with respect to “important” personal decisions, “far from emphasizing the seriousness of the decision before them, detracted both from the seriousness of the decision and the Commonwealth’s burden of proof.” We said further that “we think the better course is that all references to examples taken from the jurors’ lives should be avoided.” Finally, in Commonwealth v. Garcia, 379 Mass. 422, 438-442 (1980), considering a charge “very similar” to that in the Ferreira case, we held that the error was of constitutional dimension, and that the Ferreira decision should be applied retroactively, although “we will scrutinize more carefully jury instructions given after the date of Ferreira.”

(2) In Commonwealth v. Bjorkman, 364 Mass. 297, 307-309 (1973), we upheld a charge that a reasonable doubt is a “doubt based upon a reason,” but only after reviewing cases in which appellate courts had expressed disapproval of such definitions. Considering the charge in that case “in its entirety,” we thought that the phrase objected to, “even if of doubtful propriety standing alone, could not have caused any prejudicial shifting of the Commonwealth’s burden of proof in the minds of the jurors.” See Commonwealth v. Coleman, 366 Mass. 705, 712 (1975). Contrast Dunn v. Perrin, 570 F.2d 21, 23-24 (1st Cir.), cert. denied, 437 U.S. 910 (1978) (“good and sufficient reason” improper; reversi*600ble when followed by statement that reasonable doubt means “strong and abiding conviction”); Bumpus v. Gunter, 452 F. Supp. 1060, 1061-1062 (D. Mass. 1978).

(3) In Commonwealth v. Williams, 378 Mass. 217, 233-235 (1979), as in the present case, the judge quoted from Commonwealth v. Madeiros, 255 Mass. 304, 307 (1926), and we were persuaded “that if warnings about the consequences of misapplication of the law flow all in one direction, they might cause the jury to overcompensate and misapply the law in the other direction.” See Bumpus v. Gunter, 452 F. Supp. 1060, 1063-1064 (D. Mass. 1978). Considering the charge in the Williams case as a whole, however, we held that there was no error.

Each of the defendant’s three claims has force in the present case. The judge began his discussion of reasonable doubt with a statement that the presumption of innocence “goes with these defendants right into the jury room as you deliberate, until such time as you collectively are convinced that the Commonwealth has met” its burden of proving the guilt of the defendant beyond a reasonable doubt. He then defined reasonable doubt as “a doubt for which you can give a good reason,” and added that, “as a practical matter, that means that if you are in the jury room deliberating and you are able to stand up and face your fellow jurors and say, ‘I have a doubt and this is the good reason for it,’ then you have a reasonable doubt.” He then distinguished between “absolute proof beyond all imaginary doubt” and “a deep and abiding conviction of the defendant’s guilt amounting to a moral certainty,” and read the passages from the Madeiros case that are criticized as one-sided. Moral certitude, he said, is “the kind of certitude that the reasonable man would have about those matters of highest and most important events of his life.” After describing a heart surgery situation, he said that “clearly such an individual would look carefully into the alternatives. Would obtain all the information before he came to a decision as to what he should do. And members of the jury, you are in a similar situation. You are called upon to make a very important decision. And *601clearly you are called upon to examine carefully all the evidence that you have heard and come to a decision of which you have a moral certitude.”

We have approved the use of “moral certainty” and the use of “doubt based upon a reason” in this context, and it seems doubtful that the addition of “good” to “reason” or the dramatic phrase “stand up and face your fellow jurors” would be fatal, if those defects stood alone. Similarly, the quotation from the Madeiros case is not of itself so one-sided as to require automatic reversal. Finally, the heart surgery illustration has far less tendency to trivialize the jury’s duty than the illustrations used in the Ferreira and Garcia cases. Moreover, the judge used the heart surgery illustration more to explain the seriousness of the decision than to illustrate the required degree of certainty. Nonetheless, the cumulative effect of the defects in the charge is such that if the charge had been given after the Ferreira decision and a proper exception had been taken we would have held that there was error.

But the charge was given before the Ferreira decision, and no objection was made or exception taken. The case is not subject to the special duty imposed on us by G. L. c. 278, § 33E, in a “capital case.” Even in such a case, in Commonwealth v. Garcia, 379 Mass. 422, 441-442 (1980), we held that the error could be harmless if we were convinced beyond a reasonable doubt that it did not contribute to the guilty verdict. We are not required by decisions of the Supreme Court to overlook the failure to take exceptions. See Hankerson v. North Carolina, 432 U.S. 233, 244 n.8 (1977); Commonwealth v. Collins, 374 Mass. 596, 603-604 (1978). We would reverse the conviction in the present case only if we found “a substantial risk of a miscarriage of justice.” Commonwealth v. Freeman, 352 Mass. 556, 564 (1967). The more serious defects in the charge on reasonable doubt were disclosed in our opinions published before the charge was given, see Commonwealth v. Fluker, 377 Mass. 123, 130-131 (1979), and the fact that defense counsel made no objection suggests that they were not thought to be critical. *602The case against the defendant was strong, and we find no such risk as to require reversal.

2. Presumptions. The judge instructed the jury with respect to two presumptions: (1) “When a person by use of force enters a dwelling house in the middle of the night, it may ordinarily be presumed, in the absence of evidence to the contrary, that his intent is to steal or to do the act which he commits.” (2) “Now, possession of property taken from premises of a third party who owned the property or who had control over the property may be used by you as a presumption that the property was taken. . . . And the courts in some places talk about, if property that has been stolen is found in the possession of another person and no satisfactory explanation is made by him as to how he obtained possession of it, there is a presumption that he stole it. ” The defendant argues that the result was an impermissible shift to the defendant of the burden of proof as to his specific intent to commit larceny, citing In re Winship, 397 U.S. 358 (1970), and Sandstrom v. Montana, 442 U.S. 510 (1979). We disagree.

“The threshold inquiry in ascertaining the constitutional analysis applicable to this kind of jury instruction is to determine the nature of the presumption it describes. . . . That determination requires careful attention to the words actually spoken to the jury, . . . for whether a defendant has been accorded his constitutional rights depends upon the way in which a reasonable juror could have interpreted the instruction.” Sandstrom v. Montana, 442 U.S. 510, 514 (1979). In context it is quite clear that the judge used the word “presumption” interchangeably with the word “inference.” See Gibson v. Commonwealth, 377 Mass. 539, 542 (1979). Early in his charge he explained that an inference of a fact might follow naturally and normally, “reasonably” but “not necessarily,” from testimony to another fact, and that “you may” make all such inferences. Later, as to the intent to commit larceny, he said, “The intent that we refer to is inferred from the defendant’s conduct. It is obvious that in most cases you are not going to have any direct evi*603dence of what a person’s state of mind is. You have to determine that based upon all the facts and circumstances of which you have heard testimony during the course of a trial. Finding of intent is based generally upon that reasonably to be inferred from the conduct of the defendant.” Then came the first of the challenged “presumption” instructions. “The Commonwealth has to establish that the defendant took property. . . . That is one of the elements of the crime, and you would have to be convinced that the Commonwealth proved that beyond a reasonable doubt.” Next the judge gave the second challenged “presumption” instruction, interrupted by the following: “Now, when I say ‘presumption,’ bear in mind that I am not talking about a legal presumption. Because the Commonwealth always had the burden of establishing the defendant’s guilt beyond a reasonable doubt.” Immediately after the “presumption” charge, the judge added, “But the presumption is one of fact and not of law. Now obviously that cannot and does not change the burden which is placed upon the Commonwealth to establish the defendant’s guilt beyond a reasonable doubt.”

The quoted portions of the charge explicitly negate any shifting of the burden of proof to the defendant. No reasonable juror could have understood that the presumption was conclusive, requiring a finding of intent if the described entry or the described possession was found, or that it was rebuttable only if the defendant carried a burden of contrary proof. Thus the defendant’s claims of error fail. We adhere, however, to the view that it would be better practice to avoid the word “presumption” entirely in such situations. See Lannon v. Commonwealth, 379 Mass. 786, 793 (1980); Commonwealth v. Collins, 374 Mass. 596, 600 n.2 (1978). There is here no contention that a permissive inference of intent to steal was without rational basis in the circumstances described by the judge. See the full discussion of such inferences in Commonwealth v. Pauley, 368 Mass. 286, 292-299, appeal dismissed for want of substantial Federal question, 423 U.S. 887 (1975). Cf. Commonwealth v. Ronchetti, 333 Mass. 78, 81 (1955) (inference from forcible *604entry); Barnes v. United States, 412 U.S. 837, 839-844 (1973) (inference from possession of stolen property); County Court of Ulster County v. Allen, 442 U.S. 140, 166-167 (1979).

Judgment affirmed.

(dissenting). I am unable to agree with the majority that the charge given on reasonable doubt does not constitute reversible error. The majority, apparently recognizing the serious infirmities in the charge, state, “[T]he cumulative effect of the defects in the charge is such that if the charge had been given after the Ferreira decision and a proper exception had been taken we would have held that there was error.” Supra at 601. I take the position that reversal is required, notwithstanding the defendant’s failure to take a proper exception at his trial which occurred prior to the Ferreira decision.

This court applied the Ferreira decision retroactively in Commonwealth v. Garcia, 379 Mass. 422 (1980). The court stated, “We believe that In re Winship, 397 U.S. 358 (1970), made retroactive by Ivan V. v. New York, 407 U.S. 203 (1972), mandates retroactive application of Ferreira. We emphasize, however, that we will scrutinize more carefully jury instructions given after the date of Ferreira. Cf. Commonwealth v. Rodriguez, 370 Mass. 684, 692 (1976); Commonwealth v. Collins, 374 Mass. 596, 599 (1978); Commonwealth v. Stokes, 374 Mass. 583, 590-591 (1978). Commonwealth v. Garcia, supra at 441.

In the past when this court has given retroactive effect to a decision, we have reviewed on appeal errors at trials occurring prior to such decision, even in the absence of a proper exception at trial. See, e.g., Commonwealth v. Stokes, supra, giving retroactive effect to Mullaney v. Wilbur, 421 U.S. 684 (1975), and Commonwealth v. Rodriguez, supra. See also Commonwealth v. Collins, supra; Commonwealth v. Harris, 376 Mass. 201 (1978). In Stokes, supra at 590, *605we concluded that “with respect to trials occurring before Mullaney, a specific objection to the judge’s instructions on burden of proof need not be shown in order to secure appellate review.” In reaching this conclusion the court explicitly rejected the suggestion in Hankerson v. North Carolina, 432 U.S. 233, 244 n.8 (1977), which it now seems to embrace, namely, that with respect to pre-Mullaney cases, “[t]he States, if they wish, may be able to insulate past convictions by enforcing the normal and valid rule that failure to object to a jury instruction is a waiver of any claim of error.” See Stokes, supra at 588. The reasons given for rejecting the Hankerson approach were two-fold.

First, we noted our ‘“power to set aside a verdict . . . when a decisive matter has not been raised at trial.’ Commonwealth v. Myers, [356 Mass. 343, 347 (1969),] quoting from Commonwealth v. Conroy, 333 Mass. 751, 757 (1956),” Stokes, supra at 589. We then concluded that appellate review is warranted because “the constitutional issue of burden of proof goes to the very heart of the truth-finding function of the criminal trial and, as such, raises ‘serious questions about the accuracy of guilty verdicts.’” Id. (citations omitted). A closer examination of Commonwealth v. Myers, quoted above in Stokes, clarifies the conceptual underpinnings of this first rationale. In Myers, supra at 347, we stated, “We are mindful that, apart from § 33E, this court ‘in appropriate instances . . . has and will exercise the power to set aside a verdict or finding in order to prevent a miscarriage of justice when a decisive matter has not been raised at the trial.’ Commonwealth v. Conroy, 333 Mass. 751, 757 [1956]. This power has been sparingly used. ‘The test is whether there is a substantial risk of a miscarriage of justice.’ Commonwealth v. Freeman, 352 Mass. 556, 564 [1967].” It becomes apparent from Myers, that our conclusion in Stokes rests on our view that erroneous instructions on the burden of proof raise such “serious questions about the accuracy of guilty verdicts” as to raise “a substantial risk of a miscarriage of justice.” Hence, a defendant who received an erroneous instruction on the bur*606den of proof in a pr e-Mullaney trial and failed to take a proper exception, need not demonstrate, as a preliminary matter, that the instruction created a substantial risk of a miscarriage of justice because the court has already recognized this risk in its decision to apply Mullaney retroactively.1

Our second reason for rejecting the Hankerson suggestion rested on the inconsistency of holding “on the one hand that a substantive rule of constitutional dimension is completely retroactive and [insisting], on the other hand, that defense counsel must have anticipated the rule in the form of an objection or exception before it may be applied retroactively.” Commonwealth v. Stokes, supra at 589-590. This court has never required “clairvoyance on the part of defense counsel.” Id. at 588.

While holding in Stokes that Mullaney would be applied retroactively in the absence of an exception at trial, we also recognized the need for a less stringent standard of review in pr e-Mullaney cases. We stated, “[T]his court will bring greater expectations, and consequently more careful scrutiny of the judge’s charge as to these issues, in any case where the trial occurred after the date of Mullaney, and particularly after the date of Rodriguez.” Id. at 591. See Collins, supra at 599. Thus our decision to apply Mullaney retroac*607tively in the absence of a proper exception at trial was balanced by a realistic concern that reversal not follow automatically where the instructions given were fair, but not in technical compliance with the requirements of Mullaney. We do not demand clairvoyance on the part of a trial judge. We do require that a charge, in its entirety, place the burden on the Commonwealth to prove guilt beyond a reasonable doubt.

I find no justification for treating the retroactive application of Ferreira any differently from our retroactive application of Mullaney and Rodriguez. An erroneous instruction on reasonable doubt also raises “serious questions about the accuracy of guilty verdicts.” The charge in the case at bar trivializes the Commonwealth’s burden of proof and offends the constitutional requirements of Mullaney and In re Winship, supra.2 In light of our reasoning in Stokes, I cannot agree with the majority’s reliance on Hankerson v. North Carolina, supra at 244, n.8, for the proposition that “we are not required by decisions of the Supreme Court to overlook the failure to take exceptions.” The Hankerson approach was squarely rejected in Stokes, supra at 588. It is my position that when this court takes the dramatic step of giving a decision retroactive effect, we make a policy judgment that a substantial risk of a miscarriage of justice is involved. In my view it is inconsistent to hold that Ferreira will be given retroactive effect, and to then conclude that the instant charge, which fails to meet the requirements of Ferreira, does not require reversal because “no such risk” exists, supra at 602. This conceptual inconsistency is exacerbated by the more obvious illogic of requiring “clairvoyance on the part of defense counsel” in order for a defendant to benefit from the retroactive application of Ferreira.

*608The majority also seem to rely on the harmless error doctrine in reaching its decision that reversal is not required here, supra at 601. For the reasons stated in my dissent in Garcia, supra at 445, I do not believe that an erroneous charge on reasonable doubt can constitute harmless error. See also Connolly v. Commonwealth, 377 Mass. 527, 538 n. 16 (1979). Even assuming, arguendo, the propriety of applying the harmless error doctrine in this context, the case at bar cannot support a finding of harmless error. As the court stated in Garcia, supra at 441, “The reasonable doubt standard is most crucial in cases where central facts (such as identity, or the occurrence of an event) are at issue, and credibility plays a key role.” The court, in Garcia, supported its finding of harmless error on the ground that the evidence of guilt was “overwhelming.” Id. at 442. The present case involves no such overwhelming evidence of guilt; credibility played a key role.3

Reviewing the charge under the less stringent standard appropriate in a pre-Ferreira case, I conclude that the Commonwealth’s burden of proving guilt beyond a reasonable doubt was seriously diminished by the charge given. Accordingly, I would reverse the conviction and order a new trial.