OPINION

Appellant, Herbert A. DeSacia, was convicted by a jury on October 4, 1968, of the manslaughter of Reynaldo E. Evangelista, as charged in count II of a two count indictment. He was, at the same time, found not guilty on the first count of the indictment, which charged the manslaughter of Eugene E. Hogan. After the jury rendered its verdict, DeSacia moved for judgment notwithstanding the verdict, arguing that the verdict of conviction in count II was inconsistent with the acquittal in count I. The trial court denied the motion for judgment notwithstanding the verdict and on the basis of this denial, DeSacia appealed.

The facts relevant to this appeal center about an automobile accident which occurred in Fairbanks on September 1, 1967. On that date, shortly before midnight, appellant was driving his car, a 1959 Ford, westward along a stretch of First Avenue which is bordered to the North by the Chena River. With appellant in his car were two passengers. Ahead of the De-Sacia car was a red Ford Ranchero, a vehicle similar in design to a pickup truck, driven by Eugene E. Hogan. Accompanying Hogan in the front of the Ranchero were three passengers, one of whom was Reynaldo Evangelista; there were two additional passengers in the back of Hogan’s Ranchero. A short distance behind the De-Sacia car was a Chevrolet driven by Steve Weltz, who was accompanied by one passenger. The three cars continued to drive west along First Avenue in this order until they came to a left-hand curve. At this time, apparently, DeSacia pulled his car out to the left in an attempt to pass the Ranchero driven by Hogan; Steve Weltz remained some distance behind in his car. As the two lead cars went around the curve, side by side, traveling at between 40-60 miles per hour, Hogan lost control of the Ranchero and drove off the road into the river. Both Hogan and Evangelista were trapped in the car of the Rancero and *371killed. The four other passengers in the Hogan car managed to escape.

DeSacia was later arrested and charged with two counts of violation of AS 11.15.-040,1 the Alaska manslaughter statute; in the first count appellant was charged with the manslaughter of Eugene E. Hogan, the driver of the Ranchero, while in the second he was charged with the death of Reynaldo E. Evangelista. On September 30, 1968, a jury trial on the alleged violations of AS 11.15.040 was commenced in Fairbanks. On October 4, the jury returned verdicts finding DeSacia not guilty of the manslaughter of Eugene E. Hogan as charged in count I, and guilty of the manslaughter of Reynaldo E. Evangelista as charged in count II.

The appellant first contends that the evidence at the trial below was insufficient to allow his conviction, since it did not show that he forced Hogan’s Ran-chero to lose control and go off of the road. At the outset we must note that it is evident upon examination of the indictment and upon a reading of the trial court’s instructions to the jury that the violations charged were based on a criminal negligence theory2 Such a theory is clearly within the purview of AS 11.15.040. Thus, in order to sustain its burden of proof, the prosecution needed only to show that Hogan’s car left the road as a direct result of DeSacia’s culpably negligent handling of his own vehicle. There was no need for the prosecution to show that DeSacia intentionally forced the Hogan vehicle off of the road.

In reviewing the record of the trial below pursuant to an allegation of insufficiency of the evidence, we must view the evidence and the inferences to be drawn therefrom in a light most favorable to the state.3 The proper question on appeal is whether the finding of guilt is supported by “such relevant evidence which is adequate to support a conclusion by a reasonable mind that there was no reasonable doubt as to appellant's guilt.” 4

At the trial, testimony of witnesses established that DeSacia’s automobile attempted to pass the Hogan vehicle while on a curve; both cars were traveling at a high rate of speed on a graveled road, at night. The testimony would further support the *372conclusion that, in attempting to pass, De-Sacia did not allow the Hogan vehicle enough room to negotiate the curve; that the DeSacia vehicle commenced to slide in the direction of the Hogan vehicle; and that, as a result, Hogan was forced to swerve off of the road. While the appellant attempted to refute the prosecution’s theory at trial by introducing testimony to show that Hogan’s Ranchero went out of control and left the road when Hogan shifted gears and thereby lost traction, testimony of witnesses for the prosecution tended to show that Hogan did not, in fact, change gears or accelerate while the two cars were negotiating the corner where the accident occurred.

We are well aware that, in order to establish culpable negligence for the purposes of our manslaughter provision, a degree of conduct more reckless and wanton than would be involved in ordinary negligence is required. As we have already noted, the trial court was also aware of this distinction between ordinary and criminal negligence, and properly instructed the jury to this effect. However, we are nevertheless of the opinion that, when viewed in a light most favorable to the state, the evidence in the record is sufficient to allow the appellant’s conviction to stand.

In his next assignment of error, DeSacia argues that the jury’s verdict convicting him of the manslaughter of Reynaldo E. Evangelista is irreconcilably inconsistent with its verdict acquitting him of the manslaughter of Eugene E. Hogan. Appellant’s argument rests on the assertion that the elements of the crimes charged in counts I and II of the indictment are in all respects identical. Thus it is contended on appeal that the verdict of guilty on count II of the indictment, in light of the verdict of not guilty on count I, is not only inconsistent, but is so contradictory as to be ridiculous. Appellant would have this court reverse his conviction on the grounds of this inconsistency.

We are at the outset confronted by two arguments advanced by the state. The state contends first that the appellant has waived his right to question the consistency of the verdicts by failing to make a timely motion for acquittal before moving for a judgment notwithstanding the verdict.5 Rule 29(a) of the Alaska Rules of Criminal Procedure provides that a motion for acquittal can be made at two junctures in the course of trial: (1) at the close of the state’s evidence or (2) at the close of the evidence of defense. From this we may conclude that such a motion, if made after the close of the defense’s case, would not be timely. Thus, according to the state’s point of view, to question the consistency of the verdicts in this case, DeSacia should first have been required to move for acquittal no later than at the close of his presentation of evidence.

The flaw in the state’s reasoning on this point is all too obvious. If the state were to have its way, we would, in effect, be requiring a defendant to move for acquittal on the basis of inconsistent jury ver-*373diets before those verdicts were returned and before the inconsistency existed. Such a result would indeed be anomalous, and cannot be approved by this court. We feel, moreover, that a requirement in this instance of a motion for acquittal would not be consistent with the views we expressed in Shafer v. State, 456 P.2d 466, 467-468 (Alaska 1969).

The state next contends that even if the verdicts below were inconsistent, De-Sacia is foreclosed from raising the issue because he failed to object to the instructions of the trial court which permitted the inconsistencies to arise. Even in states where inconsistency is held to be a ground for reversal of convictions, the state argues, an objection to trial court instructions permitting inconsistency is required before an appeal will be allowed. The cases cited by the state in support of this contention are not persuasive.6 Insofar as these cases are pertinent, they stand only for the proposition that, where a defendant has been shown to be guilty by the overwhelming weight of the evidence, or where a trial court instruction specifically permits inconsistent verdicts, an appeal on the basis of inconsistency will not be heard, absent a timely objection to the trial court’s instructions. In the instant case we are confronted with neither situation. First, we cannot say that the overwhelming weight of the evidence adduced at the trial below pointed to DeSacia’s guilt. Second, the trial court instruction here in question, although allowing sufficient latitude for an inconsistent jury verdict to be returned, certainly did not specifically direct that such a finding would be acceptable.7 Even if we accept the proposition that the appellant should have objected to Instruction No. 10, it does not necessarily follow that we cannot consider the question of inconsistency on appeal. Criminal Rule 47(b) provides :

Plain errors or defects affecting substantial rights may be noticed although they were not brought to the attention of the court.

We interpreted the above provision in the case of Hammonds v. State, 442 P.2d 39 (Alaska 1968): “The meaning of Crim.R. 47(b) is that we may consider questions raised for the first time on appeal if necessary to effect substantial justice or prevent the denial of fundamental rights.” 442 P.2d at 43. Accordingly, we hold that, in the circumstances of the case before us now, Criminal Rule 47(b) permits us to consider the question of inconsistency on appeal.8

We turn next to the question of whether the conviction on count II of the indictment is necessarily inconsistent with the acquittal on count I. Upon a close scrutiny of the circumstances underlying the indictment, the conclusion seems inescapable that the two verdicts are, in fact, irreconcilably in conflict. We note first that the crime charged in the two indictments is based in each instance on the same alleged conduct of DeSacia, namely his criminally negligent operation of a motor vehicle. Since both Eugene Hogan and *374Reynaldo Evangelista were riding in the same car when that car was allegedly forced off of the road by the DeSacia vehicle, there is no conceivable way in which appellant’s conduct toward Eugene Hogan could be found to differ from his conduct toward Reynaldo Evangelista. Moreover, while it might have been possible to distinguish between the intent of the accused toward one or the other of the victims had the crime charged been based on intentional conduct, because the crime of which De-Sacia was accused stemmed from his alleged negligence toward the vehicle in which both of the victims were riding, it is virtually impossible to maintain that De-Sacia was more negligent toward one or the other of the victims.

Appellee insists, however, that the two verdicts are not in fact inconsistent, since different evidence was presented at trial to prove each of the two counts. Appel-lee points to the testimony of Dr. Raymond Evans to support its claim. Dr. Evans testified at the trial that he had performed autopsies on both victims of the accident. He stated that Reynaldo Evangelista, while trapped inside the cab of the Ranchero, vomited (apparently as a result of injuries suffered to his thorax and abdomen), aspired the vomit and asphyxiated. Dr. Evans further testified that a skull fracture rendered Eugene Hogan unconscious while submerged, causing him to drown.

Although the state is correct in asserting that evidence was presented to the effect that the cause of death differed in the case of each of the victims, it must be noted that the medical cause of death, under these circumstances, has no bearing whatsoever - on the elements of the crime charged. There can be no doubt here that both victims died as a proximate result of the automobile accident in which they were involved. It follows, then, that the difference in medical cause of death is of no legal significance.

It is obvious that the two counts of the indictment in this case charged DeSacia with identical conduct and with the same element of negligence. They differed only in that a different person was named as the victim in each instance. Under these circumstances, the verdict convicting DeSacia must of necessity be construed to be inconsistent with the verdict acquitting him.

The question of whether inconsistent jury verdicts of the kind with which we are today confronted should bp considered sufficiently prejudicial to the rights of the accused to constitute grounds for reversal is entirely another matter. This is an issue upon which courts of other jurisdictions are split.9 Although there are numerous cases, both state and federal, dealing with the issues presented by inconsistent jury verdicts,10 very few cases actually deal with circumstances similar to those of the instant case, where the appellant was charged in both counts of the indictment with crimes involving identical elements of conduct and volition. In short, very few cases deal with verdicts as inconsistent as those before us now.

*375Certainly, if we followed the federal rule set out in Dunn v. United States, 284 U.S. 390, 52 S.Ct. 189, 76 L.Ed. 356 (1932), as it is applied in the federal courts today,11 we would be compelled to hold that such inconsistencies do not provide a basis for reversal. But the question is one of first impression in the State of Alaska; hence we must decide whether to follow the federal rule, or whether to hold, with a minority of our sister states, that a truly inconsistent jury verdict will necessitate reversal.

A proper understanding of the arguments supporting the federal rule on inconsistent jury verdicts is a necessary prerequisite to our decision of this issue; such an understanding can be gained only through recourse to the origins of the federal rule in Dtmn. Initially it should be noted that prior to Dtmn, federal courts split on the matter of inconsistency. The 3rd Circuit12 followed the 8th Circuit13 in holding that inconsistent jury verdicts on multiple count indictments could furnish an adequate basis for overturning a conviction. On the other hand, the 2nd Circuit,14 the 6th Circuit,15 and the 7th Circuit16 all held that such inconsistent verdicts would not invalidate an otherwise valid conviction. In 1932 the split among the federal courts was ended by the Dunn case, in which the United States Supreme Court, in an opinion written by Mr. Justice Holmes, held: “Consistency in the verdict is not necessary. Each count in an indictment is regarded as if it was a separate indictment.” 284 U.S. at 393, 52 S.Ct. at 190, 76 L.Ed. at 358-359.

The rule stated in Dunn was supported by two propositions. Primary importance was given the argument that, had each of the counts involved been charged on separate indictments, an acquittal as to one would not have given rise to a valid plea of res judicata as to the other. From this it was concluded that it would be unjust to allow a defendant greater rights simply because the charges against him were joined in a single indictment. Despite the possibility that this rationale may have been persuasive at the time Dunn was decided, subsequent development in the law of res judi-cata has made it amply clear that Dunn can no longer be supported on the basis of such an argument.17

*376The second argument advanced to support the holding in the Dunn case is discernible in Mr. Justice Holmes’ reference to Steckler v. United States, 7 F.2d 59, 60 (2d Cir. 1925):

As was said in Steckler v. United States, (citation omitted):

‘The most that can be said in such cases is that the verdict shows that either in the acquittal or the conviction the jury did not speak their real conclusions, but that does not show that they were not convinced of the defendant’s guilt. We interpret the acquittal as no more than their assumption of a power which they had no right to exercise, but to which they zvere disposed through lenity.’ (Emphasis supplied.)

284 U.S. at 393, 52 S.Ct. at 190, 76 L.Ed. at 359. This argument has subsequently been explained and clarified by the Second Circuit Court of Appeals in United States v. Maybury, 274 F.2d 899 (2d Cir. 1960); indeed, the Maybury case makes it clear that the federal rule on inconsistent jury verdicts can be supported only by this second rationale advanced in Dunn.18 Briefly stated, the argument may be paraphrased in the following manner. Since their very origins, juries have had the power, if not the right, to act irrationally. In the context of inconsistent jury verdicts, the traditional power of irrationality is essential, since it permits a jury to express its desire to act leniently toward certain defendants by convicting on only one of several counts. Without such power, a jury would have no assurance that a desire on its part to have an accused treated leniently would be heeded by the court.

The point must be stressed that this argument, the basis for the federal rule on inconsistent jury verdicts, relies upon its stated assumption that an inconsistent verdict vyill almost certainly reflect the jury’s disposition to favor the accused by acting leniently in his behalf. The assumption has most recently been restated in United States v. Carbone, 378 F.2d 420 (2d Cir. 1967):

The very fact that the jury may have acquitted of one or more counts in a multi-count indictment because of a belief that the counts on which it was [has, sic] convicted will provide sufficient punishment forbids allowing the acquittal to upset or even to affect the simultaneous conviction. * * * It is true, as both Judge Hand and Mr. Justice Holmes recog*377nized, that allowing inconsistent verdicts in criminal trials runs the risk that an occasional conviction may have been the result of compromise. But the advantage of leaving the jury free to exercise its historic power of lenity has been correctly thought to outweigh that danger. (Citations omitted.)

378 F.2d at 422-423.19

With this assumption we are unable to agree. Although the assertion that juries have traditionally not been required to render consistent verdicts is itself subject to some doubt,20 our disagreement with the logic underlying the federal rule on inconsistent jury verdicts is not of a historical nature. Rather, we can see no basis to assume, as was assumed in Dunn, that inconsistent verdicts are the product of a jury’s disposition toward treating the accused leniently; nor can we see a basis for assuming that, in allowing inconsistent jury verdicts in criminal trials to stand, we run only “the risk that an occasional conviction may have been the result of compromise.” The truth is simply that we do not know, nor do we have any way of telling, how many inconsistent verdicts are attributable to feelings of leniency, to compromise, or, for that matter, to outright confusion pn the part of the jury.

An excellent illustration of this point can be found in the instant case. A study of the record of DeSacia’s trial below reveals that, at some time after the jury' had been charged and sent out to deliberate, the court called the jurors back to determine what their position was at that time. It is not clear from the record how long the jury had been out before it was first called back. At any rate, it is certain that at that particular time the jury was disposed six to five, with one undecided, as to the first count; no recent consideration had even been given to the second count. The jury was again sent out. At about 11:30 p. m., the court once again called back the jury to determine whether any progress was being made in reaching a verdict. With the exception of a single person all of the jurors felt that no progress had been made toward a solution.21

The hour was late and the trial judge was obviously perplexed. He stated that because there were several conventions in town at the time, no accommodations could be made for the jury on that particular night. He also said that it was, of course, out of the question to allow the jurors to return to their homes for the night. It was finally decided, after a conference with counsel in the anteroom, to let the jury vote as to whether it wanted to continue deliberations for a while longer. A vote was taken and ten members of the jury voted to continue through the night; the other two were apparently willing to accede to the wishes of the majority. The court agreed, and the jury was again sent out to deliberate. The jury continued to deliberate into the night for an undetermined length of time. Sealed verdicts of not guilty as to count I and guilty as to count II were rendered by 10:00 a. m., the following morning.

It is manifest that this particular set of circumstances suggests the possibility that the jury below, after a prolonged period of stress, reached its conclusions through compromise. Yet it is still possible that the jury acquitted the appellant on count I of the indictment only because it felt that a *378conviction on count II would be sufficient punishment. As a matter of fact it might be possible to speculate indefinitely as to what paths the jury followed in reaching its conclusions in this case. The point is that we have no reliable way to discover what really lies behind the inconsistent verdicts; and any conclusions — any assumptions on our part — would not be warranted.

What we do know, however, is that the verdict of the jury convicting DeSacia on count II of the indictment is inconsistent, and necessarily so, with the acquittal in count I. This result is irrational. In our system of criminal procedure, both at trial and on appeal, the entire focus of the fact-finding effort is upon a reasonable solution of factual issues. Accordingly, the language of our cases centers upon the element of reason. For example a jury will not be allowed to convict a person accused in a criminal matter unless it finds guilt to be established beyond a reasonable doubt. A trial court judge must enter a judgment of acquittal if he determines that fair-minded men cannot reasonably differ on the question of whether there is a reasonable doubt as to the defendant’s guilt.22 We can conceive of no reason why, under these circumstances, we cannot require at the very least, a minimal degree of reasonableness in the rendering of jury verdicts.

We could, of course, seek to distinguish the Dunn case from the case now before us. A recent case of a New York court, People v. Bullís, 30 A.D.2d 470, 294 N.Y.S.2d 331 (N.Y. 1968), involving a situation very close to’that of the instant case, reversed a conviction on the grounds that it was repugnant to, rather than merely inconsistent with, a simultaneous acquittal in another count. Such an approach, however, tends to confuse substance with semantics. In reaching our conclusion, we find it necessary only to agree with the words of Justice Butler, dissenting in Dunn:

One accused in different counts of an indictment of the same crime, there being no difference in the means alleged to have been employed, may not be adjudged guilty on a verdict of conviction on one count and of acquittal on the other. (Citations omitted.)

284 U.S. at 402, 52 S.Ct. at 194, 76 L.Ed. at 363. The state argues, in this regard, that we would be asking too much of juries in criminal matters by requiring that they render, in all cases, strictly consistent verdicts. To this we reply that, in holding as we do today, we require only that verdicts not be strictly inconsistent. Appellant’s conviction is reversed.

Having reversed appellant’s conviction, we are next faced with the question of proper disposition of the case against the appellant. First, as to the jury’s verdict of acquital on count I of the indictment below, we conclude that a retrial is precluded by the double jeopardy clause of the fifth amendment of the United States Constitution,23 and by the similar clause of the Alaska . Constitution.24 In the recent case of Benton v. Maryland, 395 U.S. 784, 89 S.Ct. 2056, 23 L.Ed.2d 707 (1969), the United States Supreme Court overruled its previous holding in Palko v. Connecticut, 302 U.S. 319, 58 S.Ct. 149, 82 L.Ed. 288 (1937), and made applicable to the states, through the fourteenth amendment, the double jeopardy clause of the fifth amendment. Accordingly, the case of Green v. United States, 355 U.S. 184, 78 S.Ct. 221, 2 L.Ed.2d 199 (1957), speaks directly to the issue with which we are confronted in the case at hand, and necessarily precludes a retrial of count I.

Certainly the argument might be made that DeSacia, by seeking a reversal of his conviction on the basis of incon*379sistency, casts doubt not only upon the validity of his conviction, but also upon the validity of his acquittal. Thus it could be contended that DeSacia waived his right to protection by the double jeopardy clause not only as to his conviction but also as to his acquittal.25 It was this very argument, however, which was confronted and resolved in the Green case. In an opinion written by Mr. Justice Black, the Court noted that any finding that a person appealing a conviction must waive his right to a plea of former jeopardy on a count of acquittal necessarily forces the person to choose between accepting his conviction or appealing at the risk of incurring an additional punishment. The Court further stated:

The law should not, and in our judgment does not, place the defendant in such an incredible dilemma. Conditioning an appeal of one offense on a coerced surrender of a valid plea of former jeopardy on another offense exacts a forfeiture in plain conflict with the constitutional bar against double jeopardy. (Emphasis added.)

355 U.S. at 193-194, 78 S.Ct. at 227, 2 L.Ed. 2d at 208. Certainly DeSacia, prior to appealing his conviction, had a valid plea of former jeopardy as to his acquittal on count I.26 Following the Green case, as we must, we cannot hold that by appealing his conviction DeSacia was forced to surrender his right to be protected from retrial on the acquittal.

We must next decide whether DeSacia may be retried on the reversed conviction. The issue which we must face here is one of collateral estoppel.27 The doctrine of collateral estoppel must be considered because of the unique situation which would inevitably arise on a retrial of DeSacia’s reversed conviction.

As we have previously stated, the elements of conduct and intent charged in both counts of the indictment below were identical. Thus, in order to secure a conviction on retrial, the state would have to adduce essentially the same evidence which was ruled upon when DeSacia was acquitted on count I. Because of this situation, it could be argued that collateral estoppel precludes retrial of the conviction we have reversed. The question that arises is whether such a plea of collateral estoppel would be valid.

There can be little doubt that the doctrine of collateral estoppel applies to the criminal as well as the civil law in Alaska. In Dapcevich v. State, 360 P.2d 789, 792-793 (Alaska 1961), we held that collateral estoppel acts, by precluding the state from litigating matters previously ruled upon by a jury, to prevent harassment by successive prosecution in criminal cases.28

In the federal law, collateral estoppel has long been held applicable to criminal cases by the United States Supreme Court. In *380Sealfon v. United States, 332 U.S. 575, 68 S.Ct. 237, 92 L.Ed. 180 (1948), it was held:

[Collateral estoppel] applies to criminal as well as civil proceedings and operates to conclude those matters in issue which the [earlier] verdict determined though the offenses be different. (Citations omitted)

332 U.S. at 578, 68 S.Ct. at 239, 92 L.Ed. at 184.29 Most recently, the United States - Supreme Court has held that, in light of Benton v. Maryland, supra, the rule of collateral estoppel is embodied in the fifth amendment as an adjunct to the constitutional guarantee of double jeopardy. Ashe v. Swenson, 397 U.S. 436, 90 S.Ct. 1189, 25 L.Ed.2d 469 (April 6, 1970). In considering the significance of collateral estoppel in criminal cases, Mr. Justice Stewart, writing for the majority of the Court in Ashe noted:

“Collateral estoppel” is an awkward phrase, but it stands for an extremely important principle in our adversary system of justice. It means simply that when an issue of ultimate fact has once been determined by a valid and final judgment, that issue cannot again be litigated between the same parties in any future lawsuit.

397 U.S. at 443, 90 S.Ct. at 1194.

The difficulty in the instant case is unlike the usual collateral estoppel problem in criminal cases, where it must be decided what particular issues a prior general verdict of acquittal resolved.30 In this case we must determine whether, because of its inconsistency, DeSacia’s acquittal on count I resolved any issues whatsoever.

In making this determination we are guided by the admonition of the Court in A she:

The federal decisions have made clear that the rule of collateral estoppel in criminal cases is not to be applied with the hypertechnical and archaic approach of a 19th century pleading book, but with realism and rationality.

397 U.S. at 444, 90 S.Ct. at 1194. Accordingly, we are well aware that óur decision as to whether DeSacia’s acquittal constitutes a meaningful adjudication of any ultimate issues of fact must be based upon an examination of the entire record, and upon a consideration of all other circumstances relevant to the appeal.31 Here we feel that the logical inconsistency of the acquittal with the conviction is obviously a factor relevant to the decision of what issues were decided by the acquittal.

We have held that DeSacia’s acquittal must be allowed to stand insofar as double jeopardy precludes a retrial on that count. But our holding in this regard should not be taken as an expression of faith on our part in the merit of the acquittal as a resolution of the issues involved in the charge. As an ingredient of the guarantee against double jeopardy, the rule of collateral estoppel requires a degree of confidence in a prior acquittal as an adjudication of ultimate issues of fact. It should be quite obvious that the inconsistency of the verdicts here reflects as much doubt upon the acquittal as it did upon the conviction. We must decide whether this element of doubt will be sufficient to preclude a claim of collateral estoppel on a retrial of the second count.

*381We have found little case law, either federal or state, to assist ns in making this determination.32 In the absence of persuasive authority, our approach is to seek a resolution grounded in considerations of fairness to both parties involved. As for the appellant, we are at a loss to discern how he would be unfairly disadvantaged by the granting of a retrial. Clearly this is not a case like Dapcevich v. State, supra, or Ashe v. Swenson, supra, where further prosecution would subject the accused to harassment by repeated and unnecessary prosecution. We fail to see how a retrial here could possibly subject the appellant to a greater chance of harassment than he would be subjected to in any other case ordered for retrial following a successful appeal.

On the other hand, if we allow the acquittal in this case, despite its doubtful nature, to be the source of a valid claim of collateral estoppel, then the state would be deprived, through no fault of its own, of the opportunity to have a full and fair resolution of the charges it has brought. Both Dapcevich and Ashe, in dealing with the problem of harassment of accused individuals, in effect encourage the prosecution to consolidate into one indictment the various charges upon which it seeks to convict. To allow a plea of collateral estoppel in this case would, we think, serve only to frustrate attempts by the prosecution to comply with the spirit of Dapcevich and Ashe.

In bringing this appeal, the appellant has successfully argued that the inconsistency between verdicts rendered his conviction meaningless. In our opinion, it would impose an unjustifiably harsh restriction on the state to allow the appellant to argue, in the same breath, that his acquittal is of sufficient certainty to suit the purposes of collateral estoppel. There is simply no reason here to force the prosecution to run a perilous gauntlet of criminal procedure. Under these circumstances, to- allow the appellant to invoke the rule of collateral estoppel would, as Judge Friendly stated in Maybury, “convert the guarantee of double jeopardy from a shield into a sword.” 274 F.2d at 905.

Accordingly, we order a new trial on the appellant’s reversed conviction, and further hold that such a retrial will not be precluded by a claim of collateral estoppel.

(dissenting in part).

Appellant was tried for manslaughter in connection with the deaths of two persons, Reynaldo Evangelista and Eugene Hogan. The evidence showed that both persons died after appellant had caused the car in which they were riding to drive off a road into a river. A jury found appellant guilty of manslaughter as to Evangelista, but not guilty as to Hogan.

Both deaths resulted from one act of criminal negligence on appellant’s part. He was either responsible for both deaths or not responsible for either. He could not rationally be held accountable for only one death and not for the other. The two conflicting verdicts were obviously and totally inconsistent. It is because of the inconsistency that this court sets aside the conviction of appellant for the manslaughter of Evangelista.

Up to this point I agree with the court. But I am unable to see why the not guilty verdict as to the death of Hogan should not *382also be set aside. The reason for disregarding the guilty verdict — and the only reason — is because it is inconsistent with the not guilty verdict. ' The overall result is irrational. Both verdicts, and not just the guilty one, are affected by the inconsistency and irrationality, so there is no way of knowing what the jury actually decided.

The court feels that it is obliged to give effect to the not guilty verdict, when it gives none to the guilty verdict, because of the double jeopardy provision of the constitution. At this point I disagree with my colleagues.

The principle of double jeopardy prevents the government from prosecuting a person more than once for the same offense. It reflects a sound policy — to prevent the government, with all its resources and power, from harassing an accused by successive prosecutions so as to afford the government a more favorable opportunity to convict.1 The double jeopardy clause “stands as a constitutional barrier against possible tyranny by the overzealous prosecutor.” 2

In order to effectuate the policy of the double jeopardy clause, it is held that a verdict of acquittal is conclusive even though it may appear to be erroneous.3 We are not concerned here with error on the part of the jury. Our concern is with the inability to ascertain, because of inconsistent verdicts, what the jury decided as to the guilt or innocence of appellant. We simply do not know what their decision was, since appellant cannot be guilty of one manslaughter and not guilty of the other.

It is also held that one appealing from a conviction of one offense may not be required, as a condition of appealing, to surrender his defense of double jeopardy as to another offense for which he had been acquitted.4 No such condition would be imposed here by setting aside the not guilty verdict, as well as the guilty verdict, and requiring a new trial as to both counts of the indictment. The court says that it cannot express faith “in the merit of the acquittal as a resolution of the issues involved in the charge.” It states that “[i]t should be quite obvious that the inconsistency of the verdicts here reflects as much doubt upon the acquittal as it did upon the conviction.” I believe what the court is saying is that the not guilty verdict is as meaningless as the court holds the guilty verdict to be. This being so, there is nothing for double jeopardy to operate upon. Appellant would not be required to waive a defense of double jeopardy because it has no application. In fact, he is not even required to appeal the conviction in order to obtain a new trial. The trial court, when faced with the totally inconsistent verdicts, ought to have disregarded both verdicts, discharged the jury and ordered another trial as to both counts of the indictment.

Because of the inconsistency, I can reach no other logical conclusion than that the jury failed to perform its function as a jury. The result is no different than if the jury had been unable to reach any verdict at all. This situation is equivalent to a hung jury — a jury unable to agree on a verdict. In such a situation double jeopardy is not a bar to another trial. As the United States Supreme Court has said:

There may be unforeseeable circumstances that arise during a trial making its completion impossible, such as the failure of a jury to agree on a verdict. In such event the purpose of law to protect society from those guilty of crimes frequently would be frustrated by denying *383courts power to put the defendant to trial again. [Emphasis added.] 5

A jury unable to agree is a “classic example” where double jeopardy has no application.6

I believe there should be a new trial on both counts ^of the indictment. As a practical matter, it probably will make no difference whether appellant is tried on both counts or only one, for under the evidence a finding of guilt or innocence as to one would require the same finding as to the other. The point I wish to make is that if the court is going to require reason and consistency of a jury, as it does in this case, it should not set a different standard for itself and ignore reason and consistency in its own decisions. That is what it does here by holding that the guilty verdict is without effect because of inconsistency with the not guilty verdict, and then holding that the not guilty verdict is effective despite the same inconsistency.