OPINION OF THE COURT ON REMAND
WOLD, Senior Judge:Appellant was tried on 25 August 1982. Pursuant to his pleas of guilty, he was convicted of larceny, a violation of Article 121, Uniform Code of Military Justice, 10 U.S.C. § 921 (1982). Sentence was imposed by the officer members of the court-martial and approved by Major General Thurman E. Anderson, who had also referred the charge to trial. In United States v. Montesinos, SPCM 18720 (ACMR 26 Oct. 1983) (unpub.), we affirmed the findings of guilty and the sentence. In pleadings to the United States Court of Military Appeals, appellant for the first time requested relief based on alleged unlawful command influence. The Court of Military Appeals remanded the case to us for further review.
Appellant contends, inter alia, that General Anderson was disqualified to refer the charges to trial and to perform the initial review of the results, that appellant was deprived of favorable character witnesses and thereby denied a fair sentencing proceeding, and that the sentencing proceedings were also infected by unlawful command influence on the members. The resolution of these matters is controlled by United States v. Anderson, 21 M.J. 670 (A.C.M.R.1985); United States v. Scott, 20 M.J. 1012 (A.C.M.R.1985); United States v. Cruz, 20 M.J. 873 (A.C.M.R.1985) (en banc); United States v. Treakle, 18 M.J. 646 (A.C.M.R.1984) (en banc), pet granted, 20 *651M.J. 131 (C.M.A.1985), and other relevant precedents.
We are satisfied that General Anderson was not disqualified to refer appellant’s case to trial. See United States v. Treakle, 18 M.J. at 654-655. Since appellant’s case includes essentially the same supporting evidence which was before us in United States v. Anderson, we hold that appellant has failed to raise the issue of deprivation of favorable character witnesses. (If there is any further evidence on this matter, appellant may request permission to submit it in connection with a request for reconsideration. See Rule 20, CMR Rules of Practice and Procedure.)1
As to appellant’s allegation that his sentencing proceedings were infected by unlawful command influence, we have no specific evidence regarding what the members in appellant’s trial heard, understood, or did. We have only the general evidence which we considered in Treakle concerning the effects of General Anderson’s actions upon members. Thus, we begin with what was said by this court sitting en banc in United States v. Treakle:
Among those who heard General Anderson speak was a member of appellant’s court-martial. His perception of General Anderson’s message was:
[Y]ou will not go to court and state that a [convicted soldier] is a good soldier; no matter how good the soldier’s performance was prior to the court-martial, you would not state that “you would like to have this soldier remain in the Army,” and that you would be willing to serve with this soldier in combat
... [W]hen considered from the perspective of a court-martial member, the logical inference [from this particular interpretation of General Anderson’s comments] is that evidence of the accused’s good duty performance and soldierly qualities should be discounted and that retention of a convicted accused is inappropriate. We conclude that the member in question drew that inference and understood it to be General Anderson’s policy for court-martial members. By the time appellant’s case was tried, the general’s message had been widely disseminated within the 3d Armored Division, but no remedial measures had been undertaken. This circumstance creates at least the appearance that the other members of appellant’s court-martial who were assigned to the 3d Armored Division had also heard the general’s message, understood him to be discouraging favorable character testimony, and drew the same inference.
As noted above, when such unlawful pressure has or appears to have been brought to bear on a member of a court-martial, the law presumes that the member was in fact influenced.
18 M.J. at 658 (footnotes omitted) (emphasis added).
As we pointed out in United States v. Anderson, this portion of Treakle “fell prey to [a] fallacy repudiated by Cruz, i.e., the premise that actual unlawful command influence and the appearance of unlawful command influence are to be treated as interchangeable factors in the unlawful command influence equation.” 21 M.J. at 678.2 As a result, in Treakle we did not determine whether there was actual unlawful command influence on court-martial members from the 3d Armored Division as a result of General Anderson's actions. Cruz makes it plain that we must address this question before we can know the nature of the problem confronting us and proceed from there to the proper solution. See 20 M.J. at 889-92.
*652Of course, in the case at bar there is no danger that any member in fact discounted favorable character testimony, since no such evidence was introduced by appellant.3 The real questions are whether any of the members at appellant’s trial in fact heard General Anderson’s message and, if so, whether they in fact understood it to be General Anderson’s policy that retention of a convicted soldier is inappropriate.
The evidence before us is sufficient to raise these issues concerning the actual effect of General Anderson’s actions, since by the time of appellant’s trial, General Anderson was well into his series of lectures and no remedial action had been undertaken. In a similar situation we found that “a reasonable person could conclude that at least one member of the [company-level] chain of command of every 3d Armored Division soldier believed that General Anderson did not want favorable character testimony for accused soldiers.” United States v. Anderson, 21 M.J. 670. What reasonable people could conclude about the chain of command through company level is also reasonable to believe about the members of courts-martial, given the makeup of General Anderson’s audiences and the sources from which members of courts-martial are drawn.4 Thus, a reasonable person could conclude that at least one member of appellant’s court-martial believed that it was General Anderson’s policy that retention of a convicted soldier is inappropriate.
In addition, we find this evidence sufficiently credible to trigger the presumption that such members complied with their commander’s perceived desires. Cf. Cruz, 20 M.J. at 887 (“[Credible evidence that a person ... reasonably understood that [a] commander had told him not to testify would, unless rebutted by the government, trigger a presumption that the [person] had complied with the commander’s order.”).
Confronted with a similar situation in Treakle, we responded as follows:
The Government has argued that we should order a hearing to allow further litigation of this issue. At such a hearing the parties would presumably explore such relevant questions as: (1) whether the other members had heard General Anderson’s message and, if so, what they understood the general’s policies to be; (2) whether any member recalled or considered these policies during appellant’s trial; (3) whether any member in fact drew the inference described above; (4) whether General Anderson’s policies were discussed by the members during deliberations or at any time during appellant’s trial; and, (5) whether any member was actually influenced in deciding appellant’s sentence.
In the case at bar, we have answers to enough of the relevant questions to render such a hearing superfluous. The member in question has affirmed that his perception of General Anderson’s policies was discussed by the members during appellant’s trial. Affidavits from other members corroborate this information. The sentence must be set aside.
18 M.J. at 659 (footnote omitted).
The case before us differs in that we do not have evidence confirming the intrusion of General Anderson’s policies into the deliberations on sentencing. We also lack evidence of the effect, if any, of General Anderson’s actions on the sentence itself. The issue of a prejudicial effect is fairly raised by the evidence in this case, but we have insufficient evidence to resolve that issue. We deem it appropriate in the circumstances of this case to order an evidentiary hearing.5 At this hearing the *653government will have the burden of persuading the court that appellant’s sentence was not prejudicially affected by unlawful command influence on the members to impose a discharge.6 The government will bear this burden because a member’s vote on sentence is necessarily “relevant to some material aspect of appellant’s case” and because unlawful influence on a member’s vote on sentence would, in the case at bar, necessarily have “caused substantial harm.” See Cruz, 20 M.J. at 887. (On the other hand, where influence on potential witnesses is involved, appellant must “[show] that the evidence in question was relevant to some material aspect of appellant’s case and that its absence caused material harm” in order to shift the burden of persuasion. Id. at 888.)
Accordingly, the record of trial is returned to The Judge Advocate General for submission to a different convening authority who will order a limited evidentiary hearing pursuant to United States v. Ray, 43 C.M.R. 171 (C.M.A.1971). At this hearing, the military judge will receive evidence and make findings of fact on the following questions:
(1) whether the members at appellant’s trial heard General Anderson’s message and, if so, what they understood it to be;
(2) whether any member recalled or considered these policies during appellant’s trial;
(3) whether any member drew the inference that General Anderson maintained a standing policy against the retention of soldiers convicted by court: martial;
(4) whether General Anderson’s policies were discussed by the members during deliberations or at any time during appellant’s trial; and
(5) whether any member was actually influenced in deciding appellant’s sentence.
See Treakle, 18 M.J. at 659. After the military judge has heard the evidence and made findings of fact, the record of the proceedings will be returned to this court for our consideration of the command influence issue. Alternatively, if the convening authority determines that such a hearing is impracticable, he may order a rehearing on sentence.
Judge FELDER concurs.. Appellant does not contend that his plea of guilty was affected by General Anderson’s actions. Our examination of the evidence of record convinces us that the validity of appellant’s plea of guilty remains unaffected.
. In the words of the court in Cruz, "the appearance of unlawful command influence is not merely a lower threshold for raising the issue ... or for finding the existence of unlawful command influence.” 20 M.J. at 883.
. We have already addressed and disposed of appellant's claim that General Anderson’s actions deprived appellant of such testimony.
. As in Anderson, our conclusion is specifically limited to favorable character evidence. 21 M.J. at 675 n. 7. Contra United States v. Mitchell, 19 M.J. 905, 906-8 (A.C.M.R.1985).
. This may not always be the appropriate response. Cf. United States v. Davenport, 17 M.J. 242, 245 (C.M.A.1984); United States v. Fort, 36 C.M.R. 242, 245 (C.M.A.1966) (a predisposition against the retention of soldiers who stand convicted of certain crimes "is practically unavoidable."). We express no opinion now.
. One of the ways the government might fulfill its burden is by showing that even if a particular member did perceive unlawful pressure to vote for a discharge, he refused to allow that pressure to influence his vote. In order to prove that fact, the government would have to "produce (or point out already existing) 'clear and positive’ evidence." Cruz, 20 M.J. at 887; see Treakle, 18 M.J. at 657, 658 n. 31.