OPINION OF THE COURT
SPISAK, Judge:What should have been a straight-forward case of fraternization, adultery, and conduct unbecoming an officer, became a complex, lengthy, and often confusing testament to how not to conduct criminal investigations and prepare courts-martial for trial. Numerous allegations of unlawful command influence and interference with defense witnesses resulted in extensive judicial hearings, over 1700 pages of transcript and 30 volumes in the record of trial (17 dedicated to exhibits alone), and a total of four investigations under Article 32, UCMJ, 10 U.S.C. § 832.
The appellant, a special agent for the Air Force Office of Special Investigations (AFO-SI), was tried by general court-martial on charges of conduct unbecoming an officer,1 indecent assault, adultery, fraternization, violating a lawful general order, sodomy, impeding an Article 32 investigation, and attempting to influence the testimony of a witness. He pled not guilty to all charges. After convicting him of one specification each of adultery and fraternization with Airman A, the court sentenced the appellant to a dismissal and 30 days confinement.
The appellant contends that the evidence is factually insufficient to support the findings of guilty; that the military judge erred in refusing to grant an evidentiary hearing on the admissibility of an exculpatory polygraph; that court members may have improperly considered extraneous information; that the sentence is inappropriate; and, that the case was so permeated with unlawful command influence that the findings and sentence should be set aside.2 We agree with the appellant’s contentions on the polygraph and unlawful command influence and set aside the findings and sentence.
I. ADMISSIBILITY OF AN EXCULPATORY POLYGRAPH
Appellate defense counsel argues that the military judge erred in denying their request for an evidentiary hearing to establish the admissibility of an exculpatory polygraph. Although the military judge made his decision without the benefit of the United States Court of Appeals for the Armed Forces’ (USCAAF) decision in United States v. Scheffer, 44 M.J. 442 (C.M.A.1996), cert. granted, — U.S. -, 117 S.Ct. 1817, 137 L.Ed.2d 1026 (May 19, 1997), we find that Scheffer requires an evidentiary hearing in the appellant’s case.
The appellant testified in his own defense, denied having sexual intercourse with Airman A, and was cross-examined on this point by the trial counsel. Under these circumstances we find that his credibility was squarely at issue and the military judge should have held an evidentiary hearing to determine whether or not the polygraph examination was sufficiently reliable to present to the members for their consideration.
Appellate government counsel’s argument that we should ignore the precedent set *774by our senior court in Scheffer is not well taken. The argument is based largely on a misunderstanding of the holdings in United States v. Kraffa, 11 M.J. 453 (C.M.A.1981), and United States v. Sparks, 18 C.M.R. 77, 1955 WL 3286 (1955). In Kraffa, the Court of Military Appeals, now USCAAF, noted that decisions of the service courts of criminal appeals are not final until the time for filing an appeal with USCAAF has passed. As a result, such decisions are inchoate and do not serve as mandates to the convening authority in the case. Neither the Kraffa nor Sparks decisions, however, say or even imply that such decisions are not prece-dential in nature or that they are to be ignored with impunity by trial judges and trial practitioners pending the outcome of appeal to the higher court. We do not believe that the inchoate nature of our decisions deprives them of validity or value to our lower courts any more than we believe an appeal to the Supreme Court invalidates a decision of USCAAF.
In addition, the government’s argument ignores the clear notice given by the US-CAAF in United States v. Allbery, 44 M.J. 226 (1996), and reiterated in United States v. Kelly, 45 M.J. 259 (1996). In Allbery we were told that “[i]n the absence of a superseding statute or an intervening decision of [USCAAF] or the Supreme Court of the United States, [USCAAF precedent cases are] absolutely binding on the Court of Criminal Appeals.” Allbery, 44 M.J. at 228 (emphasis added). Our choice, should we disagree with that precedent, is to either distinguish the present case or urge reconsideration by our senior court. Id. Because the Supreme Court recently heard oral argument in United States v. Scheffer, we see neither a need to distinguish the present case nor to urge reconsideration by US-CAAF.
II. UNLAWFUL COMMAND INFLUENCE
A. Background
The appellant contends that numerous acts and comments by commanders, supervisors, investigators, and legal advisors involved in this case resulted in a pattern of unlawful command influence which requires corrective action by this Court. Specifically, he alleges that (1) the AFOSI (“Command” for this case) unlawfully endeavored to influence and did influence the testimony of key witnesses; (2) the AFOSI and legal office interfered with defense access to witnesses and evidence; and (3) military members were treated adversely as a result of their status as defense witnesses.
At the first pretrial Article 32 investigation, more than a year after the events about which he testified, Captain V, a defense witness, said that during a camping trip with the appellant, he (Captain V) called Airman A asking if she or anyone else from the command post was coming out for the camping trip. The AFOSI agents investigating the case subsequently interrogated Captain V twice under rights advisement as a suspect for making false official statements because they believed the appellant had called Airman A not Captain V. See Article 31, UCMJ, 10 U.S.C. § 831. Major Bergan, the lead case agent, and Special Agent Cash, the appellant’s commander and commander of the AFOSI detachment at Malmstrom Air Force Base (AFB), both testified that either the Staff Judge Advocate (SJA) or one of the other lawyers requested the first reinterview of Captain V and said that it should be under rights advisement. However, the lawyers remembered only asking that Captain Y be re-interviewed, not that he be advised of his rights under Article 31.
During the second interrogation of Captain V, two AFOSI agents came to his home while he was on convalescent leave and asked him to return to their office with them. One of the agents rode in Captain Ys car, while the second followed in the agents’ car. During the interrogation one of the AFOSI agents referred to the Captain as “Mister” V and told him that the appellant was “going down” and that Captain Ys career would be over if he continued “the way” he had been going. Captain V thought the agents were trying to influence him to change his statements. This second interrogation was purportedly part of an investigation of Captain Ys participation in an office sports pool with a maximum value of $60.00. However, according to Cap*775tain Vs testimony, very little was asked about the “pool” during the interrogation. This is supported by the agent notes for these interviews which contain no information related to the sports pool. The Wing Commander later offered Captain V nonjudicial punishment under Article 15, UCMJ, 10 U.S.C.A. § 815, for the sports pool, but withdrew that action and gave him a letter of reprimand (LOR). A senior officer in Captain Vs supervisory chain told him the LOR was delayed because they didn’t know if his disciplinary action could be separated from the appellant’s case. Another superior officer told Captain V his disciplinary problems were related to “something bigger,” but didn’t explain what. Although that officer later testified that by “something bigger” he meant Captain Vs career, the Captain understood him to mean the appellant’s case.
Although 16 people in the command post participated in the sports pool which Captain V organized, the only other people who were punished for this activity were the appellant and Technical Sergeant B, another defense witness. Additionally, court-martial charges were preferred against Master Sergeant D, another defense witness, for misuse of a government computer system and his involvement in the sports pool. However, the sports pool charges were later dropped.
Senior Airman McG gave the appellant a written statement in which he stated that during one winter camping trip with the appellant, at which he and the appellant were the only campers, they called the command post to see who else might be coming out. Airman A answered the phone and agreed to join them after work. This statement was written 23 months after the camping trip. Ever the hawks for detail, the AFOSI investigators determined Airman McG’s statement was false because phone records indicated that Airman A was called at home. Therefore, like Captain V, they interrogated Airman McG under rights advisement for making a false official statement. This interrogation took place just two days before Airman McG was to travel to Malmstrom AFB from Iceland to testify as a defense witness in appellant’s court-martial. During this interrogation, Airman McG revealed for the first time that the appellant had told him that he (the appellant) had “slept with” Airman A. Airman McG had never before implicated the appellant in any wrongdoing with Airman A. In fact, in his written statement in support of the appellant, Airman McG said about the camping trip on which the appellant supposedly “slept with” Airman A, “Never did I witness or hear anything come from either Amn [A] or [the appellant]. It would have been impossible for [the appellant] to commit such an act, for I was only a foot away from Amn [A]. Personally, I do not think [the appellant] would put his career on the line like that.”
As a result of Airman McG’s statement and the fact that the appellant had called him several times asking for a statement, the agents asked Airman McG to call the appellant and let them listen to the call. They also asked him to have the appellant meet him on his arrival in Montana and to wear a “wire” so they could record this meeting. However, the agents needed permission from the Air Force General Counsel to conduct this “wire” surveillance. Because the appellant was represented by counsel and the trial was scheduled to begin in just eight days, the General Counsel denied the request. The AFOSI agents then requested “emergency” approval on the grounds that Airman McG was about to travel to Malmstrom and there was insufficient time to make a second request to the General Counsel. No other facts or circumstances had changed since the prior rejection of this request, but the “emergency” request was approved by AFOSI command after coordination with the 20th Air Force SJA, the General Court Martial Convening Authority’s legal advisor.
After the meeting, the AFOSI transcribed the conversation between the appellant and Airman McG. That transcript was considered by the investigating officer at an Article 32 investigation into additional charges, including obstruction of justice. During preliminary hearings in the case, the military judge determined that the transcript and, therefore, the evidence discussed in the Article 32 investigating officer’s report were inaccurate and misleading. Specifically, the transcript stated that the appellant told Air*776man McG, “All you got to say is that I don’t know,” when in fact the appellant had said, “All you got to say is the truth.”3 The only explanation offered by the trial counsel to explain this strange conflict in the transcript was that there were numerous errors in the transcript “in part ... just due to the nature of the tape and the location and the background music, and the taping equipment.” However, trial counsel presented no evidence on this glaring discrepancy.
Yet another defense witness, Staff Sergeant M, testified that after Special Agent Cash interviewed him, she told him not to talk to the defense counsel about the case. The defense counsel called Sergeant M the next day, and when he was unable to reach Special Agent Cash to give her his written statement, Sergeant M took the statement to the defense counsel. At trial Special Agent Cash denied telling Sergeant M not to talk to the defense counsel, but said she may have told him he didn’t need to give anyone a statement at that point.
In March, 1995, Sergeant D and two others from the command post were told by a lieutenant colonel in their chain of supervision, that they “need to stay away from Plumb. He’s trouble, and you don’t need any of that rubbing off on you.” Similarly, when Lieutenant K, who later became a defense witness, asked her commander for leave to attend the court-martial, he told her to “stay away” from the appellant’s case because it was “getting ugly.”
Two other defense witnesses had worked as AFOSI sources. One, Mr. K, a former enlisted member, was “terminated” as a source, according to internal AFOSI documents, because he was still associating with the appellant and had “even testified on behalf of [the appellant] and as a result, [Mr K’s] integrity and especially his honesty [was] very much in question.” The other, Senior Airman S, was terminated without being told why or even that she had been terminated.4
When Senior Airman H, another defense witness, asked his commander to remove him from the control roster early, his commander refused, in part because he had been “seen in the gym talking to [the appellant].” As a result, Airman H was unable to test for promotion.
In addition to these numerous acts and statements directed toward defense witnesses, the record indicates that the lead AFOSI investigator in this case, Major Ber-gan, went skiing with a court-member before the trial knowing that the member had been selected to serve on the appellant’s court-martial. Even more disturbing, however, are the threats this same Major Bergan made toward one of the defense counsel during trial. During a break in the proceedings, Major Bergan told the defense counsel, “I’m going to talk to ... my buddies in the region and we’re going to take care of you ... I know where you live____ Do you ever drive on Randolph [AFB, Texas]?” (Both the AFOSI region to which Major Bergan had been assigned before his retirement, and the Central Circuit Trial Judiciary office to which the defense counsel was assigned are located on Randolph AFB).
Defense counsel also complained at trial and on appeal that they were denied access to witnesses and evidence. Specifically, Major Bergan, who was by then retired, refused to allow defense counsel to review the AFO-SI Report of Investigation which, for some inexplicable reason, he had in his possession during trial. Defense also complained that when an AFOSI agent began interviewing witnesses who might provide favorable character evidence on behalf of the appellant, Major Bergan ordered the agent to stop and did not include any of the interviews or witness names in the report of investigation.5
*777Finally, although not raised by appellate defense counsel, we also note that while two defense witnesses were advised of their rights for making false official statements concerning a minor event that occurred one or two years earlier, the same fate did not befall government witnesses whose initial statements contained apparent errors. Rather, when the AFOSI suspected that Cadet B had given an inaccurate statement, related to an incident that had occurred only seven months earlier, he was re-interviewed without being advised of his rights or told that he was suspected of making a false official statement. Yet another government witness, Airman C, lied to investigators in order to protect Airman A. However, there is no evidence that when she was interviewed a second time that she was read her rights for making a false official statement. This disparity of treatment, seen in relation to the administration of punishments for the “sports pool,” is not adequately explained in the record of trial. However, two notable differences between the false statements from Cadet B and Airman C, and those of Captain V and Airman McG are readily apparent. First, Cadet B and Airman C were government witnesses. Second, the error in their statements related not to acts of the appellant, but to misconduct by the complainant and an AFOSI agent. It is apparent that the investigating AFOSI agents sought to exonerate their fellow agent by getting Cadet B to correct his statement. The record is silent, however, on why these same agents faded to advise Airman C of her Article 32, UCMJ, rights.
B. Discussion
As appellate courts have repeated on numerous occasions, unlawful command influence is the mortal enemy of military justice. United States v. Thomas, 22 M.J. 388, 393 (C.M.A.1986), cert, denied, 479 U.S. 1085, 107 S.Ct. 1289, 94 L.Ed.2d 146 (1987). The standard of review of claims of unlawful command influence requires the appellant to (1) “allege sufficient facts which, if true, constitute unlawful command influence;” (2) show that the proceedings were unfair; and (3) show that the unlawful command influence was the proximate cause of the unfairness. United States v. Stombaugh, 40 M.J. 208 (C.M.A.1994), cert, denied, 513 U.S. 1156, 115 S.Ct. 1113, 130 L.Ed.2d 1077, (1995) (quoting United States v. Levite, 25 M.J. 334, 341 (C.M.A.1987) (Cox, J., concurring)). The threshold triggering further inquiry into an allegation of unlawful command influence is low, however, there must be more than bare allegations and speculation. United States v. Johnston, 39 M.J. 242, 244 (C.M.A.1994). “[T]here must be something more than an appearance of evil to justify action by an appellate court in a particular case. ‘Proof in the air, so to speak, will not do.’” United States v. Allen, 33 M.J. 209, 212 (C.M.A. 1991), cert, denied, 503 U.S. 936, 112 S.Ct. 1473, 117 L.Ed.2d 617 (1992) (quoting Pollock, Torts, 455 (11th ed.)).
Once the defense has made a color-able showing of unlawful command influence, the burden shifts to the government to either disprove the existence of unlawful command influence or to prove that it did not effect the proceedings. United States v. Ayala, 43 M.J. 296, 299 (1995). Exactly how heavy a burden the government faces has not been clarified by USCAAF. United States v. Gerlich, 45 M.J. 309, 310 (1996). However, the Navy-Marine Court of Criminal Appeals has applied the clear and convincing standard to the initial question of whether command influence exists. United States v. Lawson, 33 M.J. 946, 951 (N.M.C.M.R.1991), affd on other grounds, 36 M.J. 415 (C.M.A.1993). We concur with our Navy brethren for a number of reasons.
As noted in Thomas, when it interferes with an accused’s access to witnesses or is directed against counsel, unlawful command influence infringes upon Sixth Amendment protections of the U.S. Constitution. When directed toward court-members, it threatens the accused’s right to a fair and impartial forum. Thomas, 22 M.J. at 393. Constitutional violations of this sort are prejudicial *778unless a reviewing court concludes beyond a reasonable doubt that the trial result was not affected. Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967). However, if the government is able to demonstrate that no unlawful command influence occurred, there can have been no such constitutional violation and the need for a beyond a reasonable doubt analysis is foreclosed. This does not, however, mean that the burden faced by the government should be reduced to the preponderance of the evidence test required by R.C.M. 905(c). Rather, because the threat to fairness is so great, the government should be held to the higher standard of providing clear and convincing evidence to overcome this first obstacle. If they succeed, no further action by the military judge or appellate courts will be necessary because they will have disproved the existence of unlawful command influence.
If, however, the government fails to meet this initial burden, we hold that the military judge must take corrective action unless he is convinced beyond a reasonable doubt that the unlawful command influence did not prejudice the proceedings. Thus, the government must either prove by clear and convincing evidence that there was no unlawfid command influence, or, failing that, it must prove beyond a reasonable doubt that the unlawful command influence did not prejudice the proceedings. This higher burden comports with that placed on appellate courts which may not properly affirm either the findings or sentence unless persuaded beyond a reasonable doubt that the results have not been affected by unlawful command influence. Thomas, 22 M. J. at 394.
Interference with defense access to witnesses, discouraging witnesses from testifying, attempting to influence court members, and threatening officers of the court are all forms of unlawful command influence. Stombaugh; United States v. Dykes, 38 M.J. 270 (C.M.A.1993); United States v. Zagar, 18 C.M.R. 34, 1955 WL 3279 (1955); United States v. Hunter, 13 C.M.R. 53, 1953 WL 2390 (1953). These and several other acts which raise the ugly specter of command influence are present in this case. When such acts are committed by commanders, the appropriateness of the appellation “unlawful command influence” is clear. However, where the complained of acts are performed not by commanders, but by others acting either as investigators or on their own (i.e., officers telling potential witnesses to “stay away” from the case), it may be less obvious that unlawful command influence or interference is afoot.
While the actions of criminal investigators such as the AFOSI agents involved here, do not normally implicate commanders, we find in this case that they do. Here the AFOSI investigation originated as an internal investigation against the appellant, an AFOSI special agent. Therefore, each of the AFOSI investigators was acting under the mantle of the AFOSI command authority. Moreover, Special Agent Cash was the appellant’s commander, thus the actions of the Malmstrom AFB AFOSI special agents who worked under her supervision and direction are imputed to her and become actions of the commander.
The lieutenant colonel who told his subordinates to avoid the appellant, however, was neither associated with the AFOSI nor the commander of either the appellant or any potential witness. Nevertheless, his statements to members of the command post that they should “stay away from” the appellant could reasonably be understood as an attempt to influence or interfere with potential witnesses. Such attempts are specifically proscribed by Article 37 when undertaken by a “person subject to” the UCMJ. As it is with claims of unlawful command influence, the burden of proof for interference with witnesses is on the party raising the issue. Stombaugh, 40 M.J. at 213. Moreover, where there is a disparity in rank or position, attempts to influence a witness may rise to the level of unlawful command influence. Id. We find that, the statements by senior officers to personnel assigned to the command post and those made to Lieutenant K, rise to this level under the facts of this case.
We, like the military judge in this case, are satisfied by the sheer weight of allegations and the testimony provided during the lengthy judicial hearings, that the *779appellant has provided far more than “bare allegations and speculation” and has, therefore, met the first prong of the Stombaugh test. We turn next to the question of fairness in the appellant’s trial. In evaluating this second prong of the Stombaugh test, the military judge relied heavily on the fact that “[e]very witness testifying ... stated without hesitation that their (sic) ability and willingness to testify truthfully has not been affected by any government action.”6 It is well settled that “such perfunctory statements from subordinates on the effects of command influence [are] inherently suspect, not because of the credibility of the witness, but because of the difficulty of the subordinate in ascertaining for himself the effect of an attempted command influence.” United States v. Rosser, 6 M.J. 267, 272 (C.M.A.1979); Zagar, 18 C.M.R. at 38; United States v. Carlson, 21 M.J. 847, 851 (A.C.M.R.1986). See also United States v. Adamiak, 15 C.M.R. 412, 419, 1954 WL 2310 (1954) (court member may experience difficulty in recognizing what has influenced him); United States v. Treakle, 18 M.J. 646, 658 (A.C.M.R.1984), affd, 23 M.J. 151 (C.M.A1986) (member’s denial he was influenced insufficient to rebut the presumption of influence).
After concluding that the defense had raised facts which, if true, constituted unlawful command influence, but had failed to demonstrate any harm to the appellant, the military judge merely asserted his belief that the standards of fairness had been and would continue to be met.7 He did not, however, take any corrective measures to prevent further interference with defense witnesses or assure defense access to witnesses and evidence. More disturbing still, the military judge failed to take any remedial actions related to the threats made to the defense counsel by Major Bergan. Although defense counsel asked only for an investigation by an outside agency and that the military judge consider these threats when deciding the motion to dismiss, the military judge had an independent duty to ensure the fairness of the proceedings and the ability of the defense counsel to represent the appellant free of threats and interference. The military judge should have taken steps to rid the trial of any possible taint. At a minimum, he could have recommended commanders’ calls in both the AFOSI detachment and the wing to inform all potential witnesses that if requested as defense witnesses, their testimony was a duty; and, ordered the government to produce all witnesses requested by the defense, then informed each of the duty to testify and assured them that no adverse actions would follow. United States v. Sullivan, 26 M.J. 442, 443 (C.M.A1988). Additionally, steps should have been taken to neutralize Major Bergan’s threats to defense counsel. While we realize that it would be an extreme sanction, we believe it would have been appropriate for the military judge to refuse to permit this Major Bergan to testify. Even though he was not called by the government in its case-in-chief, making such an order in open court would have gone a long way to restore public confidence in the fairness of this trial.
Unfortunately, instead of taking such remedial actions, the military judge relied on the willingness of witnesses to testify. This reliance dealt exclusively with the impact on this appellant and looked only for prejudice to this case. Even if we agreed that there was no actual harm to appellant, our inquiry would not be over. Rather, whenever unlawful command influence is an issue in a case, we must consider not only the effects of “actual” influence, but also whether the appearance of unlawful command influ*780ence exists. Our concern in apparent unlawful command influence cases is not only that the appellant receive a fair trial, but also that the public perceives military justice as fair and impartial. Rosser, 6 M.J. at 272; United States v. Allen, 31 M.J. 572 (N.M.C.M.R. 1990). If there is an appearance of unlawful command influence, we must then consider whether or not remedial action is required. Rosser, 6 M.J. at 271-272. We find that the military judge erred as a matter of law by failing to consider the cumulative effect the egregious and multiple forms of unlawful command influence had on the appearance of fairness and freedom from command influence required for courts-martial. United States v. Bradley, 47 M.J. 715 (A.F.Ct.Crim. App.1997); Rosser, 6 M.J. at 271-272.
While they failed to so find, our Army brethren have noted that “a case may occur in which the appearance of unlawful command influence is so aggravated and so ineradicable that no remedy short of reversal of the findings and sentence -will convince the public that the accused has been fairly tried.” United States v. Cruz, 20 M.J. 873, 892 (A.C.M.R.1985), rev’d on other grounds, 25 M.J. 326 (C.M.A.1987). We have found just such a case — a case where witnesses believed investigators were trying to influence them; where government investigators obtained “emergency” approval for a wire surveillance which had been disapproved by the Air Force General Counsel; where those same investigators prepared an inaccurate transcription of that surveillance which implicated the appellant in crimes he did not commit; where commanders and supervisors alike warned witnesses away from the trial and the appellant; where witnesses were punished or denied favorable treatment in part because they associated with the appellant or supported his defense; where government investigators denied the defense access to evidence and threatened defense counsel; where a government investigator socialized with a court-member immediately before trial; where defense witnesses were warned of their rights against self-incrimination for having made minor errors in prior statements, while one government witness was merely encouraged to reconsider his statement and another was simply re-interviewed; and where at least one witness was told not to talk to defense counsel. In our combined 90-plus years of service as Judge Advocates, we have never seen a case so fragrant with the odor of government misconduct, nor one so rife with witnesses who believed they had been the targets of command influence. Because we are unable to conclude beyond a reasonable doubt that the appellant was not prejudiced by the numerous and egregious acts of command influence and interference with witnesses, we must resolve our doubts in his favor. United States v. Johnson, 34 C.M.R. 328, 331,1964 WL 5022 (C.M.A1964); United States v. Kitchens, 31 C.M.R. 175, 180, 1961 WL 4555 (C.M.A.1961); United States v. Mamaluy, 27 C.M.R. 176, 181, 1959 WL 3587 (C.M.A.1959).
As Judge Cox noted in his concurring opinion in United States v. Levite, when we find “the existence of command influence lurking in the record of trial or allied papers of a court-martial, we shall go to extraordinary lengths to be certain ‘beyond a reasonable doubt’ that neither the findings nor the sentence have been tainted----” 25 M.J. at 341. Here the evidence of command influence is not merely “lurking in the record”— rather, there is a veritable cavalcade which overshadows a few defense witnesses taking the stand to testify that they were not influenced by it.
III. DISQUALIFICATION OF THE STAFF JUDGE ADVOCATE
While not raised on appeal, we note that Major F, the same judge advocate who authored the defective April 19, 1995, Article 34, UCMJ, pretrial advice, also authored and signed the post-trial staff judge advocate’s recommendation. Preparation of the pretrial advice does not ordinarily disqualify a SJA from participation in the post-trial review, at least if his pretrial advice was “proper in all material respects.” United States v. Collins, 6 M.J. 256, 257 (C.M.A. 1979). See also United States v. Lynch, 39 M.J. 223, 228 (C.M.A.1994); United States v. Engle, 1 M.J. 387 (C.M.A.1976). However, R.C.M. 1106 contemplates that the SJA who *781authors the post-trial recommendation will be sufficiently impartial as to provide the convening authority with a balanced and objective evaluation of the evidence. United States v. Crunk, 15 C.M.R. 290, 293, 1954 WL 2289 (1954). Where the pretrial advice misstates a material fact or arrives at an erroneous factual conclusion, the staff judge advocate is disqualified from the post-trial review. Collins, 6 M.J. at 257. Here, the convening authority’s SJA did recuse himself and his staff, including Major F, from writing the third pretrial advice after the military judge declared the April 19, 1995, advice to be defective. We commend him for doing so. However, Major F then signed the post-trial recommendation as Acting Staff Judge Advocate, totally negating the SJA’s earlier cleansing acts. While Major F’s recommendation did not discuss the prior pretrial advice or argue that it was not defective as had occurred in Engle, or enter into a factual dispute with defense counsel as occurred in Lynch, the blatant inaccuracies of the April 19, 1995, recommendation disqualified Major F from authoring the post-trial recommendation. Id.
Here, however, there is more to disqualify Major F than a defective pretrial advice. Major F also authored the first pretrial advice which characterized the appellant as being “like a shark in the waters, [who] goes after the weak and leaves the strong alone.” Some pre-trial acts or comments by a SJA may be so antithetical to the integrity of the military justice system as to disqualify the SJA from further participation. Engle, 1 M.J. at 389. We believe this language is so contrary to the integrity and fairness of the military justice system that it has no place in a pretrial advice. Having the author of those hostile words and of the defective pretrial advice sign the staff judge advocate’s post-trial recommendation to the convening authority had the effect of causing this simmering pot of unlawful command influence to boil over and extinguish the fire of fairness altogether.
IV. CONCLUSIONS
The appellant’s trial was so permeated with unlawful command influence that we are unable to say beyond a reasonable doubt that neither the findings nor sentence were tainted. Additionally, the appellant was entitled to an evidentiary hearing to determine whether or not the polygraph evidence was relevant, material, and sufficiently reliable to warrant presentation to the court members. For these reasons, the findings and sentence are set aside and the case is returned to The Judge Advocate General for referral to a new convening authority who may authorize a rehearing. Should a rehearing be ordered, we expect the military judge to take appropriate measures to purge the proceedings of any residual taint.
Chief Judge ROTHENBURG, Senior Judge PEARSON, and Judge MORGAN concur.
. Two specifications of engaging in an unprofessional relationships with two female airmen; one specification of having sexual intercourse with an airman while on duty; and, one specification of making a “wrongful and dishonorable” statement to a non-commissioned officer.
. We heard oral argument at the United States Air Force Academy on September 24, 1997.
. The military judge ruled that "material and substantial defects” in the Article 32 investigator’s report and in the SJA's pre-trial Advice required remedial correction. A continuance was granted and another Article 32 investigation was held at which a corrected transcript was presented.
. Internal AFOSI documents indicate that Senior Airman S was "terminated” as a source because she had revealed her status to someone outside AFOSI, had lied to AFOSI agents, was the victim of assault, and had not provided any useful information.
. There was also confusion over whether a government attorney had advised the primary prose*777cution witness, Airman A, not to tell defense counsel about an alleged incident of sodomy which was later charged. On this one point, the record satisfies us that Airman A misunderstood the government attorney. After defense counsel raised the matter with government counsel, well before trial, the witness did talk openly with defense counsel about all of her allegations.
. The military judge also noted that the defense counsel against whom threats had been made, expressed his willingness to continue to represent the appellant and that he would not be affected by the threats. While we believe counsel meant to represent his client fully, we are concerned that these threats did affect counsel’s actions as is evident from the fact that the defense team then changed its plans to avoid having this counsel question the offending witness.
. While not crucial to our findings, we are disturbed that the military judge’s 12 "essential findings of fact” constitute little more than a recitation of the testimony of the various witnesses. Even where there were clear conflicts in the testimony of two witnesses, the military judge made no effort to clarify which witness he found more credible or on which version of the evidence he relied. We have, therefore, reached our own conclusions. Article 66(c), UCMJ, 10 U.S.C. § 866(c).