concurring.
I agree that because no proof was adduced during defendant’s case, it was harmless error for the district court to make its finding on the admissibility of the co-conspirator hearsay declarations at the close of all the evidence. I concur and write separately, however, to express my conviction that we should no longer prefer the order of proof established in United States v. Bell, 573 F.2d 1040 (8th Cir.1978).
In Bell, this court recommended that district courts permit the government to introduce putative co-conspirator hearsay declarations into evidence subject to being “connected up” with later proof that the defendant was involved in a criminal conspiracy with the hearsay declarant. This recommendation was made as a pragmatic concession to the difficulties, expense, and delay that would arise if the government first had to prove the conspiracy before the hearsay could be admitted. Yet, as the Fifth Circuit has realized, Bell’s preferred order of proof “may be more burdensome and expensive of prosecutorial and judicial effort than a reordering of the proof.” United States v. James, 590 F.2d 575, 582 (5th Cir.) (en banc), cert. denied, 442 U.S. 917, 99 S.Ct. 2836, 61 L.Ed.2d 283 (1979). Allowing the government the leisure to connect-up the foundation to the proof at a later time substantially increases the risk that the jury will be exposed to inadmissible evidence that is highly prejudicial. In such instances a mistrial will often be required to eliminate the prejudice a cautionary instruction cannot cure, resulting in the “inevitable serious waste of time, energy and efficiency that accompanies a mistrial.” Id. at 582; United States v. Roe, 670 F.2d 956, 962 n. 2 (11th Cir.1982). See, e.g., United States v. Coffman, No. CR 81-0-10 (D.Neb. Nov. 4, 1981) (memorandum order) (where after five calendar weeks of trial, the district court excluded voluminous putative co-conspirator hearsay testimony and dismissed several of the major counts against the defendants because the “record was devoid of any competent independent evidence establishing the existence of a conspiracy.”); United States v. Payner, 590 F.2d 206, 207 (6th Cir.1979) (detailing the tortuous procedural meanderings the government must undertake in order to appeal an adverse Bell-type ruling under 18 U.S.C. § 3731). The great waste inherent in a mistrial may also cause a district court to hesitate granting a mistrial in favor of delegating to the jury the task of determining whether a conspiracy has been adequately proven. See United States v. Mas-tropieri, 685 F.2d 776, 788-89 (2d Cir.1982) (discussing the proper roles of a judge and jury in determining whether a conspiracy exists).
On balance, the dangers that inure from Bell’s preferred order of proof far outweigh its practical justifications. I favor the adoption of a hybrid of the Fifth and Tenth Circuits’ approaches to this problem. The Fifth Circuit strongly prefers that district courts conduct a pre-trial hearing on the admissibility of putative co-conspirator hearsay declarations, but permits the government to connect-up in difficult cases, subject to the district court’s finding on admissibility at the close of all the evidence. See United States v. James, 590 F.2d at 582-83. The Tenth Circuit also holds that the better practice is for the court to determine the admissibility before the hearsay is admitted. United States v. DuFriend, 691 F.2d 948, 951 (10th Cir.1982). Accord United States v. Roe, 670 F.2d at 962 n. 3 (“To be sure, a district court should not lightly forego the advantages of a James hearing.”); United States v. Jackson, 627 F.2d 1198, 1218 (D.C.Cir.1980). But in addition, the Tenth Circuit holds that the court should normally make its determination during the government’s case-in-chief. United States v. DuFriend, 691 F.2d at 951. Accord United States v. Gantt, 617 F.2d 831, 845 (D.C.Cir.1980); United States v. Ziegler, 583 F.2d 77, 80 (2d Cir.1978). I agree that the government alone should shoulder its burden of proving the defendant’s criminal complicity with the declarant. The government should not be able to rely upon the defendant’s proof to meet that burden. Nor should the defendant be put to the choice of either challenging the *271government’s foundation for the hearsay on appeal, or presenting his defense to the jury and thereby run the risk of proving the government’s case.
I would strongly recommend that district courts conduct pretrial James hearings. But if the government wishes to forego the advantages of a James hearing in a particularly difficult case, it should prove the conspiracy during its case-in-chief. The district court should then be required to make a finding on the record whether it is more likely than not that the defendant and the declarant were involved in a conspiracy and that the declarant’s hearsay statement was made in furtherance of the conspiracy. If the court finds that it is more likely than not, the defendant should be permitted to present exculpatory evidence on the admissibility issue during its case-in-chief. Finally, at the close of all the evidence, the district court should be required to make the ultimate determination of whether a conspiracy was proved by a preponderance of all the evidence.1
We have expressed a preference for pretrial hearings in the past. See United States v. Macklin, 573 F.2d 1046, 1049 n. 3 (8th Cir.1978). It has become common practice, however, to allow the government to introduce the hearsay first and lay the foundation for the hearsay’s admission into evidence later. It is my judgment that it is time to make a strong pronouncement in favor of holding pre-trial hearings to determine the admissibility of a putative co-conspirator’s hearsay declaration. Accordingly, I concur.
. At least one court has adopted the argument that the district court cannot make its finding at the close of the government’s case because a preponderance standard implicitly anticipates that the defendant’s evidence will be taken into account. See United States v. Winter, 663 F.2d 1120, 1141 (1st Cir.1981). This argument, however, does not militate against having the district court make an initial “more likely than not” finding on the record at the close of the government’s case based upon the government’s evidence and the defendant’s cross-examination, while withholding an ultimate determination until after the defendant presents his evidence during his case-in-chief. This type of procedure would have the salutory effect of limiting proof of the conspiracy to the evidence presented in the government’s case-in-chief. As a pragmatic matter this effect may not be achieved at the trial level. On appeal, however, the reviewing court will be able to scrutinize the district court’s initial finding and limit the proof establishing the conspiracy solely to the evidence adduced by the government during its case-in-chief. The appellate court can then weigh that evidence against the defendant’s proof that no conspiracy existed or that the statement was not made in furtherance of the conspiracy.
The government may contend that the above procedure is unfair because it should be given a chance to rebut the defendant’s exculpatory evidence. If the government has a legitimate concern of surprise, it should opt for a pre-trial James hearing where it may be entitled to rebuttal.