I agree with my colleague that the Court of Appeals erred reversibly in this case when it set aside the defendant’s conviction for conspiracy to commit murder. I am constrained to write separately, however, because I believe there is a different and more correct rationale for the decision we reach on both points to which my colleague has written.
I
There can be no question that Michigan is a "no one-man conspiracy” state. Our conspiracy statute is a bilateral conspiracy statute which requires proof of an agreement between two or more persons. It is not, as the people contend, a unilateral conspiracy statute that allows conviction without proof of agreement between two or more persons. The Michigan statute provides:
*36"Any person who conspires together with 1 or more persons to commit an offense prohibited by law, or to commit a legal act in an illegal manner is guilty of the crime of conspiracy punishable as provided herein”. (Emphasis added.) MCL 750.157a; MSA 28.354(1).
To interpret the foregoing statutory language as a unilateral conspiracy law would effectively write out of the statute the word "together”. Moreover, our cases have consistently held that "there can be no conspiracy without a combination of two or more persons”. People v Atley, 392 Mich 298, 310; 220 NW2d 465 (1974); People v DiLaura, 259 Mich 260, 262; 243 NW 49 (1932).
Since the statute requires proof of an agreement of at least two persons, the common-law "no one-man conspiracy” rule is applicable. That rule provides that "if two are tried together for a conspiracy in which no additional persons are implicated, a verdict finding one guilty and the other not guilty requires a judgment of acquittal of both”. See Perkins, Criminal Law (2d ed), p 622. The purpose of the rule at the common law was to prevent the enforcement of inherently defective or inconsistent verdicts in conspiracy cases. Such inconsistency resulted when, in the joint prosecution of several alleged conspirators, one co-conspirator was convicted while all of his alleged co-conspirators were acquitted. The effect analytically was that the factfinder in such cases found simultaneously that "an agreement between two or more persons” existed and that it did not exist with regard to the same alleged conspirators. At common law the rule was easily applied since co-conspirators were always tried jointly and a per se rule fulfilled the rationale underlying the common-law "no one-man conspiracy” rule, precluding inherently defective or inconsistent verdicts.
*37As my brother points out, however, it has come to be more and more commonplace that alleged co-conspirators are tried separately as in the case at bar. The question arises then whether the "no one-man conspiracy” rule is violated when one of two alleged co-conspirators is convicted following a verdict of acquittal or a directed verdict of acquittal for insufficiency of evidence of the conspiracy charge against the other co-conspirator in a previous trial. My colleague appears to conclude that the reason the "no one-man conspiracy” rule is inapplicable in the case before us is that the conspiracy charge against alleged co-conspirator Margaret Craze was dismissed by the trial judge as a matter of law on a motion for directed verdict, while defendant Anderson was later convicted of the same alleged conspiracy by a jury. In my judgment, that distinction is of no significance.
For purposes of application of the "no one-man conspiracy rule”, it matters not whether, in the separate trial of Craze, the trial judge found as a fact upon a motion for a directed verdict of not guilty that there was insufficient evidence of the conspiracy to submit the question to the jury or a jury returned a not guilty verdict in the case. The result is the same: a determination that no conspiracy existed. Whether the "no conspiracy” decision was made by the trial judge in the Craze case in ruling upon a motion addressed to the sufficiency of the evidence, as indeed it was, or by a jury, has no bearing whatever on the quality of the proof of a conspiracy in the subsequent and separate trial of Anderson.1 The reason the "no *38one-man conspiracy” rule is inapplicable to Anderson’s case is because he was not tried jointly with bis alleged co-conspirator. It is simply irrelevant that the proofs in Craze’s case were insufficient to survive a motion for a directed verdict. Anderson’s trial was separate and the quality of the evidence in that case was separately evaluated.
I conclude, therefore, that the common-law "no one-man conspiracy” rule applies to the typical common-law situation, the joint trial of alleged co-conspirators resulting in inconsistent and therefore inherently defective verdicts. It does not apply where alleged co-conspirators are separately tried because there is no inherent defect when the separate juries return different verdicts in the separate trials. As is shown in Platt v State, 143 Neb 131; 8 NW2d 849 (1943), and its progeny, different verdicts in different conspiracy trials do not impeach one another since the verdict of acquittal may be based on many factors other than a failure to establish an agreement. Similarly, the fact that a trial judge makes a finding that the prosecuting attorney produced insufficient evidence to support a finding of conspiracy in a separate case against one conspirator does not preclude a jury in a subsequent trial from concluding that the evidence of conspiracy in that case was sufficient to warrant a verdict of guilty. The rationale, however, is not that the trial judge rejected the sufficiency of the evidence in one case rather than a jury rejecting it. The inapplicability of the "no one-man conspiracy” rule arises from the fact of separate trials.
II
While I am in agreement with my colleague that *39no reversible error resulted in this case from the ex parte communication by the judge with the jury, I cannot concur fully in the rationale advanced.
My colleague suggests three reasons that the ex parte communication with the jury was not prejudicial: 1) the communication occurred before the jury deliberations began, 2) the judge expressed his intention to engage in an ex parte communication with the jury, and 3) neither party objected to the communication even though they had been put on notice by the court.
While I agree with the second and third of the foregoing points, in my judgment the first of the factors noted by my colleague is of no consequence. Whether the jury had begun deliberating or not is irrelevant to a determination that the defendant must show actual prejudice. My view is fortified, I believe, by the decisions of this Court in the three cases cited by my colleague: People v Kangas, 366 Mich 201; 113 NW2d 865 (1962), Zaitzeff v Raschke, 387 Mich 577; 198 NW2d 309 (1972), and People v Cain, 94 Mich App 644; 288 NW2d 465 (1980), rev’d 409 Mich 858 (1980) (adopting dissenting opinion of M. F. Cavanagh, 94 Mich App 647).
In Kangas, p 206, this Court declared that the defendant need not show prejudice when the ex parte communications between the judge and jury occur "after [the jurors] have retired for deliberations”. (Emphasis added.) In Zaitzeff, p 579, this Court stated that the defendant need not show prejudice when the ex parte communications between the judge and the jury involved "entering the jury room while the jurors are there”. (Emphasis added.) In Cain, the dissent which we adopted merely quoted the language of Kangas and Zait*40zeff. Consequently, the significant factor in determining whether a defendant is relieved of the obligation to show prejudice upon a claim of ex parte communications with the jury is not whether the jury has actually begun deliberations. The crucial fact is whether the jurors had "retired” and were in the jury room when the ex parte communications occurred.
In this case the jurors had indeed retired and were in the jury room when the trial judge went and spoke to the jurors. Consequently, the general rule expressed in Kangas and Zaitzeff would allow the defendant a new trial without a separate showing of prejudice. However, this is not an ordinary or general case. In this case, the trial judge, from the bench and on the record, expressed his intention to communicate with the jury outside of the presence of the parties. Both parties agreed to the procedure whereby the judge would go to the jury room, have the door opened, and instruct the jurors to begin deliberations. The purpose in prohibiting ex parte communications between a judge and the jury is to protect the parties’ rights to fair deliberations. Where, as here, the judge acted with the parties’ leave and no party or counsel for any party voiced any objection to the judge’s proposed procedure, the rationale underlying the prohibition against ex parte jury communication is not violated, and the ex parte communication does not mandate a new trial without a showing of prejudice.
There is simply no showing of prejudice in this case, and I concur with my colleague’s conclusion that a new trial is not required, but the commencement of deliberations should not be a factor in the determination. This view should not be taken in any way as approval of the practice *41employed in this case. It was an unwise and potentially prejudicial procedure. The better, simpler, and wiser practice would require the trial judge to communicate with the jury only from the bench in the courtroom and on the record. While experience suggests that that may sometimes be a cumbersome and time-consuming procedure, it is a price worth paying to preserve the integrity of the factfinding process, the appearance of jury independence, and the avoidance of the expense, delay, and consumption of judicial resources involved in appellate evaluation of jury room contact between the court and the jurors.
Reversed.
Brickley, Cavanagh, and Boyle, JJ., concurred with Ryan, J.Had Anderson and Craze been tried jointly and the case dismissed "as a matter of law” by the trial judge as to Craze, and Anderson later convicted of the conspiracy by a jury verdict, the distinction between the "legal” determination of no conspiracy and the later factual determination on the question would be of no moment in *38determining whether the "no one-man conspiracy” rule had been violated. The "no one-man conspiracy” rule would apply.