OPINION
HARDBERGER, Justice.
This is a murder case. The conviction is largely based on an eye-witness identification. The points of error are three: (1) factually insufficient evidence, (2) in-court identification and (3) excluded evidence of a civil lawsuit. All points are substantive and well-briefed by both sides.
Facts
Brenda Epperson, 24, was killed in her apartment on May 17, 1992. She was stabbed 12 times. The motive was, and is, difficult, to understand. She was neither sexually molested, nor robbed. She was well-liked and successful in her work as an insurance adjuster. Her friends were many; her enemies, if any, were unknown. Because she screamed, the time of her death can be fairly closely determined: around 3:40 a.m. at the beginning of a Sunday morning. While the hour was late, and she had been out with girlfriends that night, she had neither been drinking nor using drugs. Only a few minutes before, she had been brought home by her girlfriends, who watched her until she was safely in her locked, lighted apartment.
The witness who heard the scream and called the police shortly thereafter was Lewis Devlin, a neighbor who lived in the adjacent apartment building on the second floor. His apartment faced the third-floor apartment of Epperson. It was later to be determined that it was 87 feet, 10 inches between the apartments. Not being certain as to what he should do about the scream, and seeing nothing, he did nothing. But shortly thereafter he heard a crashing noise and he looked at Epperson’s apartment again. The apartment was lit, as was her balcony which faced him one story above his. This time he saw a tall, thin man come out of the apartment onto the balcony, look around and step over the railing on the outside edge of the balcony. The next thing he saw was a large object shoot toward the ground. Devlin looked back at the balcony, unable to believe that anyone would have voluntarily jumped the 26 feet from the balcony to the ground, but saw the balcony was now empty. Concluding correctly that the object he had seen falling was indeed the man he had seen on the balcony, Devlin called the Universal City police at 3:51 a.m. They arrived within one minute, while he was still talking to the dispatcher, and ran up the stairs to Epper-son’s front door (and the only door except the sliding entrance onto the third floor balcony). There were no signs of a forced entry. After beating on the door, they kicked it off the frame. Epperson’s lifeless [324]*324body was jammed up against the door, but they were able to push it open. There was much blood: on the door, on the floor, on the walls, on the rag, on the vertical Venetian blinds that covered the sliding door that went onto the balcony, and blood on the railing of the balcony. The sliding door was off its rail, bent outwards; the screen behind the sliding door totally knocked off. The police then went downstairs, expecting no doubt that a person having jumped three stories might still be there, or at least somewhere nearby in an injured condition. The only thing they found, however, was an indentation in the gravel surrounding the building. There was a tree and a shrub in the vicinity but no evidence that the killer had fallen into these, or that they had broken his fall. Whatever injuries might have been expected in someone falling such a great distance, it is undisputed that the killer was still mobile enough to leave the scene. No suspects were arrested that night, or for many weeks to come despite the best efforts of the Universal City police, the Texas Rangers and the science laboratories of the Department of Public Safety.
In the latter part of September though, the defendant, Ralf Loserth, was indicted by a San Antonio grand jury. Loserth, who was an Army reserve lieutenant, then on duty in Indiana, drove back to San Antonio and turned himself in. Eventually he stood trial, testified in his own behalf that he was not guilty to no avail, and was found guilty of murder. The jury sentenced him to the extraordinarily low sentence of 25 years, considering the extreme savagery of such an innocent victim with no mitigating circumstances.
Factual Insufficiency
Loserth’s first point of error is that there is factually insufficient evidence to support the verdict, and that, therefore, there must be a reversal, either with instructions to acquit, or remanded for a new trial.
The Court of Criminal Appeals has only recently decided that courts of appeals have constitutional and statutory authority to conduct factual sufficiency review in criminal eases. See Clewis v. State, 922 S.W.2d 126 (Tex.Crim.App.1996). The court first clarified that the courts of appeals do have constitutional and statutory authority to conduct factual sufficiency review in criminal cases. After having examined the evolution of appellate judicial power, the court concludes: “... from the beginning, ‘appellate jurisdiction’ included the power to examine ‘factual sufficiency,’ and further, that every appellate court with criminal jurisdiction recognized, acknowledged and utilized that power ...” Clewis, supra at 131. Further, the court held that our duty to review the facts, when properly raised, is mandatory: “When their jurisdiction to review fact questions is properly invoked, the courts of appeals cannot ignore constitutional and statutory mandates.” Id. at 128.
But while the appellate courts have the authority, and the duty, to review fact questions, great deference must be given to the jury’s findings:
“In conducting a factual sufficiency review, an appellate court reviews the factfinder’s weighing of the evidence and is authorized to disagree with the factfinder’s determination. This review, however, must be appropriately deferential so as to avoid an appellate court’s [sic] substituting its judgment for that of the jury.”
Id. at 133.
As this court has previously ruled:
“But courts of appeals should use considerable restraint in exercising their power to overturn the jury’s work. The Magna Charta forced King John to give rights to juries, not appellate courts.”
Peterson v. Reyna, 908 S.W.2d 472, 478-79 (Tex.App.—San Antonio 1995).
At the outset, it can be said that the question of insufficient evidence in this case, as presented to the jury, is a close one. Although the subsequent points of error seek to exclude some evidence that went to the jury, and include some evidence that did not, our review of factual insufficiency is done with a view as to what the jury actually heard without consideration of the two other points of error. The following facts were developed:
[325]*325The principal eyewitness, Devlin, was unable to give much of a description to the police. Both the night of the murder, as well as four days later when he gave a written statement to the police, Devlin could not do better than to say the person was (1) tall, (2) thin, and (3) wearing dark clothes like a jump suit. From then on, including several conversations with police officers and even after being hypnotized by a Texas Ranger, Devlin could not, or would not, elaborate further on this description. Two and a half months after the murder (July 27) Devlin was again called in by the Universal City police. He was shown a single color photograph of the defendant. The police explained that they did not show Devlin a traditional line-up because they were just trying to find out if the defendant had been seen around the apartment complex. Whatever the question, Devlin’s memory improved dramatically. At that moment, Devlin positively identified the defendant as the man on the balcony the night of the murder. He never again wavered, and identified Loserth as the man on the balcony (who was most certainly the murderer) in his trial testimony. Devlin’s explanation of the late identification was that he had been afraid, and that he felt the defendant had seen him and might come after him. In support of his being afraid is the fact that he had moved out of his apartment the day after the murder, though less clear is why seeing the photograph would make him less afraid. In any case, the jury heard this testimony of an eyewitness making a positive identification of the killer. Cross-examination did not reveal any doubts of the witness: he was certain.
The jury obviously believed Devlin because there was no other evidence to place Loserth at the scene of the crime. Loserth, a former acquaintance, or boyfriend of the victim, depending on the viewpoint, was extraordinarily cooperative with the police. He gave statements without an attorney, he allowed the police to search his apartment, his garage and his car. He twice gave hair samples from his head, armpit and pubic areas. He gave blood samples, saliva samples, fingerprints, and palmprints. He allowed the police to take clothes from his closet to have fabric samples tested, knives from his kitchen and bedroom to see if they would correlate to the wounds on the victim. When indicted he drove more than a thousand miles back to San Antonio to turn himself in the next day, and at trial, waived his Fifth Amendment rights and took the stand in his own behalf. None of the physical items taken by the police, or tested by them, ever linked Loserth to the crime.
There was one other eyewitness that gave corroborating evidence that Loserth was in the vicinity at the time surrounding the murder. This woman was Eileen McGraff, a district manager for the local San Antonio newspaper, who was on her way to work. Between 4:15 a.m. and 4:30 a.m., while driving 25 miles per hour, she approached a pedestrian who was walking in the same direction as she was driving. This attracted her attention as it was unusual for anyone to be out walking at that time of the morning. As she passed this man, she looked at him and he at her. She continued driving, and looked in her rearview mirror, but the man was gone. At the trial she identified the man she passed that night as Loserth. The man was walking south, about a mile from the murder scene and about a mile and a half from Loserth’s apartment. Both Loserth’s apartment and the murder scene were back to the north, so he was walking in the opposite direction. The man was not limping.
One last person that saw something was Lemuel Johnson, who also lived in the complex. He saw a “pretty tall” white man running through the apartment parking area dressed in a dark blue or black warm-up suit with a hood. The first time he saw him was at 3:15 a.m. and he was going towards the victim’s apartment, and then sometime later he saw the person coming away from the apartment, but this time he was limping. Other than the above description, he could not give any greater detail.
Loserth’s whereabouts from around 5 a.m. the morning of the murder are not in dispute. He was doing reserve duty in his capacity as a lieutenant at Fort Sam Houston. He testified he got there at 4:50 a.m., and his testimony was backed up by one of his superiors, Major Haas. He was cleanly [326]*326dressed in the appropriate fatigue uniform. His first job was to climb an 8-foot chain link-fence, with barbed wire on top, to turn on the field lights. He did this without difficulty, and came out the same way. A little later he turned them off the same way. Other reserve officers had arrived by 5:15 a.m., and confirmed Loserth was already there and working. He stayed throughout the day, voluntarily staying for a lunch cookout and volleyball games in the afternoon, in which he played. No one that day, and several people testified, noticed any limping, restriction of movement or scratches or cuts. He also didn’t act out of the ordinary, either in speech or actions. The next day, Monday, he was back at work at his regular job in the computer software business, I.Q. Software. No one noticed anything different, either in his physical, social or mental behavior. He attended the victim’s funeral on Wednesday. One of Epperson’s co-workers, Carol Gardner, said she saw him limping and rubbing the top of his leg at the funeral, but she was the sole witness to so testify. Numerous witnesses, including the police themselves, who talked to him the day after the murder said he was not limping, or injured in any way that could be observed. They also asked him to pull up his pants leg to see if they could see scratches, or bruising. He did so, but they saw nothing.
Loserth, 26, is a graduate of St. Mary’s University where he received a commission as a second lieutenant. He has no criminal record, outside this case. He was working at the time of the murder, and until the trial in his field of computers. He was doing reserve duly on the weekend of this murder. He had just completed a series of physical tests at Fort Sam the day before the murder. He was in good, but not outstanding shape. He scored 212 out of a possible 300: enough to pass, but not excel. He is 6'2" tall, and weighs around 190 pounds. He had been engaged since December, 1991, to. a University of Texas student, Marissa Lee Gayton, and was still engaged to her at the time of the death of Epperson. Loserth had met Epperson in the fall of 1991 in the building where they both worked; they also lived close together, about .6 of a mile apart with Loserth’s apartment just up the road to the north from Epperson. It was 15 to 17 miles from Loserth’s apartment to Ft. Sam, and he testified it took about 20 minutes to drive it. Other witnesses said you could do it in 15 minutes.
With the given time of the scream at 3:40 a.m., and Loserth arriving at work in an appropriate condition at 4:50 a.m., there was conflicting testimony as to whether it was physically possible for Loserth in that one hour and ten minutes to get from the murder scene to where he could clean up (there was a great deal of blood), change clothes and drive to work. This is further complicated by the McGraff testimony about a man walking a mile from the scene between 4:15 a.m. and 4:30 a.m. But juries have the duty to sort out conflicting facts and they did so on this occasion, and there is evidence to support that implied finding.
There was also conflicting testimony whether it was possible for Loserth to have jumped three stories and have been able to go anywhere very fast, much less to participate in physical activities without signs of being hurt. Here too though, the jury can weigh the facts, and there was some evidence to support them, especially with the eyewitness identification made by Devlin.
The implied motive was a jilted, and angry lover. But the testimony is not overly strong. There was no evidence of anger, arguments, violent incidents or that Epper-son was afraid of Loserth. There is some evidence that, in the fall of 1991, a romance was desired by Loserth. He wrote her two poems, the gist of which was that he wanted to spend more time with her. They would hardly be classified as passionate though, and he testified they never had sexual intercourse at any time. How much time they spent together is unclear by the evidence. The testimony ranged from two dates (Loserth’s testimony) to many times, stretching from October, 1991, until Epperson’s death. (Some of her friends.)
Loserth did not always help himself. None of these things, mentioned below, proved he committed a murder, or any crime, but they did not endear him to the jury:
[327]*327(1) When first questioned by the police as to where he was that night before he went to Ft. Sam, he said he was spending the night with his fiance, Ms. Gayton. This was a lie. A few days later, Loserth voluntarily admitted that he had made that up, and said that he was alone that night, and simply went to bed in his apartment.
(2) He understated his relationship with Ms. Epperson. He told the police he had only had a couple of dates with her, and those back in the fall of 1991. He explained he had gotten engaged in December of 1991 to Ms. Gayton, and had no further relationship with Ms. Epperson except he would occasionally run into her at work as they both worked in the same building. Several witnesses testified he continued to come by and see her at her work station, and that they would talk on the phone, well into the spring of 1992. It is not clear just how long in 1992, or whether these contacts had stopped in the weeks before her murder; it is also not clear how the relationship was terminated, or whose idea it was to terminate it, if indeed it was. It is also unknown if it was a romantic relationship. Suffice it to say, it was something more than Loserth led the police to believe.
(3) Loserth was cooperative with the police, but, of course, they didn’t always tell him everything they did. One of the unannounced things they did was to search his garbage. Nothing was ever found of any importance. On one occasion though, they found a note. This note, written by Loserth, informed them that if they kept searching through his garbage they might find Jimmy Hoffa. The police were not amused, and it’s fairly probable neither was the jury.
(4) Loserth was asked for second hair samples while he was on temporary reserve duty in Indiana. He agreed, and a Doctor Duncan was given the job of getting the hair samples. While he was doing so, and not knowing the history of the case, Duncan asked Loserth what this was all about. Lo-serth replied that some girl in San Antonio had gotten “hacked up.” Duncan was chilled by the insensitivity of the remark, and the jury, knowing more than Duncan about the relationship, were probably more so.
The State in its brief refers to the above actions and words as describing a man “disingenuous, chameleon like and extraordinarily narcissistic.” There may be evidence to support such a view.
The jury was within its rights to consider all of the evidence, including the above actions and words. In addition, they had the two eyewitnesses in-court identification of the defendant. Devlin’s testimony was especially critical as it put the defendant directly at the scene of the murder. McGrafTs testimony, while important, was essentially corroborative. It put the defendant in the general vicinity and time of the crime.
Considering the great powers of a jury in this state to decide factual matters, and being deferential to those findings, we find there was factual sufficiency to support the verdict. Point of error one is overruled.
The In-Court Identification
(Point of Error Three)
Loserth timely filed a motion to suppress identification testimony. During the trial, the court held a suppression hearing out of the presence of the jury. At thé conclusion of the hearing, the trial court suppressed Devlin’s photographic identification, but allowed him to make an in-court identification. The defendant’s third point of error complains of this action.
There is a close connection between an out-of-court photographic identification and an in-court identification. This is true because an impermissibly suggestive photographic identification can also taint the in-court identification that follows. Each case must stand on its own facts, but certain guidelines have evolved to help the courts in their evaluations. The leading case in this area is Simmons v. United States, 390 U.S. 377, 88 S.Ct. 967, 19 L.Ed.2d 1247 (1968).
“... convictions based on eyewitness identification at trial foEowing a pretrial identification by photograph wiE be set aside on that ground only if the photographic identification procedure was so impermissibly suggestive as to give rise to a very sub[328]*328stantial likelihood of irreparable misidenti-fication.”
Id. at 384, 88 S.Ct. at 971.
The court then goes on to discuss problems that can be induced by the police that can cause an incorrect identification.
This danger will be increased if the police display to the witness only the picture of a single individual who generally resembles the person he saw ... Regardless of how the initial misidentification comes about, the witness thereafter is apt to retain in his memory the image of the photograph rather than of the person actually seen, reducing the trustworthiness of subsequent lineup or courtroom identification.
Id. at 383, 88 S.Ct. at 971.
The Supreme Court in Neil v. Biggers, 409 U.S. 188, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972), set out factors for the trial court to consider:
“... the factors to be considered in evaluating the likelihood of misidentification include the opportunity of the witness to view the criminal at the time of the crime, the witness’ degree of attention, the accuracy of the witness’ prior description of the criminal, the level of certainty demonstrated by the witness at the confrontation, and the length of time between the crime and the confrontation.”
Id. at 199, 93 S.Ct. at 382.
In Manson v. Brathwaite, 432 U.S. 98, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977), the Court adopted the Biggers factors, but emphasized that “The standard, after all, is that of fairness as required by the Due Process Clause of the Fourteenth Amendment.” Id. at 113. The Court of Criminal Appeals in Madden v. State, 799 S.W.2d 683, 695 (Tex.Crim.App.1990), cert. denied, 499 U.S. 954, 111 S.Ct. 1432, 113 L.Ed.2d 483 (1991) stated that “the practice of showing a single photograph to the prosecuting witness is condemned”. But, the court continued, “If the totality of circumstances reveals no substantial likelihood of misidentification despite the suggestive identification procedure, then the identification testimony will be deemed reliable, and therefore admissible ...” Id. at 695. Similar language is found in Delk v. State, 855 S.W.2d 700 (Tex.Crim.App.1993), cert. denied; 510 U.S. 982, 114 S.Ct. 481, 126 L.Ed.2d 432 (1993), and Cantu v. State, 738 S.W.2d 249, cert. denied, 484 U.S. 872, 108 S.Ct. 203, 98 L.Ed.2d 154 (1987).
A review of the cases shows that a cléar majority of the eases to consider the question of impermissibly suggestive photographs have ultimately affirmed the conviction even though they were used as a part of the state’s case. The usual rationale is that there was a high reliability factor, and frequently that there was plenty of other evidence that the defendant committed the crime. Further, the law imposes a heavy burden on the defendant.
“The burden is on the defendant to show by clear and convincing evidence that the in-court identification is unreliable.”
Delk, supra at 706. For similar language, see Madden, 799 S.W.2d at 695.
The “impermissibly suggestive” standards are not always hollow words though and some cases are reversed when there is a substantial doubt raised about the reliability of the identification.
“... convictions based on eye-witness identification at trial following a pretrial identification by photograph will be set aside ... if the photographic identification procedure was so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification.”
Simmons, supra at 384, 88 S.Ct. at 971.
It was under this reasoning that the Fifth Circuit reversed the conviction in United States v. Sutherland, 428 F.2d 1152 (1970).
“Because the District' Court found that an impermissibly suggestive photographic identification created a substantial likelihood of misidentification, yet allowed the question of an in-eourt identification to go to the jury, we reverse and remand for a new trial.”
Id. at 1154. It appears that the trial court in Loserth did the same thing. At the very least the court’s ruling shows a substantial distrust of the out-of-court identification, enough to keep it out. Yet, the in-court identification, based on the sole photo shown to Devlin at the police station was allowed.
[329]*329Sutherland states: "... if the picture spread in the particular case was impermissi-bly suggestive either in the photographs used or the manner or number of times they are displayed,” then, If the judge makes such a determination, he then should determine if the impermissibly suggestive picture spread gives rise to a likelihood of irreparable misidentification. If both elements are found, Simmons prohibits the use of the in-eourt identification.” Id. at 1155. Also see United States v. King, 321 F.Supp. 614 (1970), for another, similar case in which the photo identification was held to be impermis-sibly suggestive and the identification was suppressed.
Dispensa v. Lynaugh, 847 F.2d 211 (5th Cir.1988) is another Fifth Circuit case arising out of a Texas conviction that held “The out-of-court identification was inadmissible and the in-court identification could not stand without it.” Id. at 221. This case, while going through the Biggers/Simmons factors, said that the most important of these factors was the witness’s inability to describe the criminal before the inadmissibly suggestive act. Speaking of the factors, the court said:
“Of these, by far the most significant in this case is the lack of accuracy of the witness’s prior description of the criminal.”
Id. at 220. The court, in that case where the crime was rape, felt the victim’s earlier inability to say whether her assailant had a moustache, was hirsute, and had a tattoo was fatal to the later identification, either in or out of court.
In our case, Devlin could never get beyond “tall and thin” for almost two and a half months. His written statements never even mention whether the criminal was Anglo, Mexican-Ameriean, or African-American. In his testimony, he explained that he was never asked, but this is hard to believe, and was disputed by the police. Common sense tells us that a hard-working police investigating unit, and there is plenty of evidence that they were hard-working, would have done everything they could have done to have a better description from the one eye-witness who put the criminal at the murder scene itself. They even tried hypnosis on Devlin in an effort to probe any suppressed subconscious memory, but this was a failure. Devlin was an Air Force policeman and certainly would have understood the need for an accurate description of a murderer. Devlin said he didn’t give a better description because he was afraid. Fear is an element that courts can consider. However, he never said he was lying before his late identification of the photo, although he did say he purposely did not give a full description. His testimony seems to be a combination of (1) he wasn’t asked, and (2) he suppressed the memory. The instant change from virtually no memory to complete certitude on being shown a single photograph, without any of the safeguards of a traditional lineup, raise very serious questions on the reliability of the identification. It was these very questions that, no doubt, caused a careful trial judge to suppress this out-of-court identification. The logical question that follows has to be: is the in-court identification more reliable than the out-of-court identification? There are circumstances where it could be. For example, a witness might have already given a fairly accurate description of the criminal, and then been shown a sole photograph and the out-of-court identification suppressed. The in-court identification would still be valid because it would be consistent with what was already known. The danger in a case such as the instant case is that it is far from clear what the witness could have identified before being shown the sole photograph.
In the case of Jimenez v. State, 787 S.W.2d 516 (Tex.App.—El Paso 1990, no pet.), which was a rape case, the case against the defendant very substantially rested on the victim’s in-court identification of the defendant as her assailant. The court first • recognized the “difficult and heavy burden” that a defendant has to demonstrate by clear and convincing evidence that the trial identification was irreparably tainted by impermissible suggestive pretrial identification procedures. Id. at 519. The court then went through the Big-gers/Simmons factors, as adopted by our Texas courts in Herrera v. State, 682 S.W.2d 313 (Tex.Crim.App.1984), and concluded, in a fact-intensive analysis, that the in-court identification was unreliable, and irreparably [330]*330tainted by pretrial identification procedures that created a substantial likelihood of mis-identification. Similar results were reached, after similar analysis of the facts, in Rawlings v. State, 720 S.W.2d 561 (Tex.App.— Austin 1986); Coleman v. State, 505 S.W.2d 878 (Tex.Crim.App.1974); and Proctor v. State, 465 S.W.2d 759 (Tex.Crim.App.1971). After reviewing all the cases, including the majority that did not reverse, and the minority of those that did, the following general conclusions can be made: (1) the defendant’s burden to overturn a conviction on a tainted in-court identification is heavy; (2) the defendant usually fails, but not always and (3) every case must be considered on its own facts, (4) using the Biggers/Simmons/Herrera nonexclusive guidelines (5) considering the totality of the circumstances. We do so:
1. The opportunity to view the criminal at the time. It was night, and Devlin was almost 30 yards away. On the other hand, the apartment was lighted, there was a porch light on the balcony, and another light on the side of the building. The time for observation was brief. The criminal walked out of the apartment, stepped over the rail, looked around and jumped. The lighting was not so good that Devlin could actually tell whether something was thrown off the balcony or the man jumped. He concluded he jumped when the man was no longer there.
2. The witness’s degree of attention. High. Devlin was specifically looking to see what was happening.
3. The accuracy of the prior description of the criminal. Virtually non-existent beyond “tall and thin”. Unable to expand the details even under hypnosis.
4. The level of certainty demonstrated at the trial confrontation. Absolute certainty, giving no ground to cross-examination.
5. The time between the crime and the confrontation. About two and one-half months from the crime to being shown the lone color photo (May 17 to July 27 when he was shown the photo; statement given on July 28.). Almost two years from the crime to the in-court identification. (May 17, 1992 to March 4,1994).
In evaluating the above factors we feel, as the court did in Dispensa, supra:
“Of these, by far the most significant in this case is the lack of accuracy of the witness’s prior description of the criminal.”
Id. at 220.
We do not overlook the testimony of Dev-lin that'says he was afraid, and we agree that fear can be a powerful, and sometimes a logical motivator. But there is simply no evidence before being shown the lone color photograph that he was capable of being any more descriptive than “tall and thin.” He testified he willfully held back some descriptive details; but the impression given was that on being shown the photo, that it brought it all back. Lighting, distance and time for observation were far from ideal: given the conditions, “tall and thin” is not unreasonable.
The police explained that they showed the photograph to Devlin to simply find out if Loserth had been seen around the apartments. The police had several other photos of similar appearing males, gathered for the purpose of a line-up which was prepared on June 8 that they did not show him. They could just as easily have asked if any of these males had been seen around the apartment complex. The police already knew from their investigation that Loserth had been around the apartments more than he originally indicated. So, it is difficult to understand the purpose of the single photograph unless it was to suggest that this was the very “tall and thin” person he had seen on the balcony. In any ease, whether it is called a line-up or some other name, we feel that it was impermissible, suggestive and resulted in a substantial likelihood of irreparable mis-identification. We agree with the trial court that the out-of-court identification was inadmissible. But we further think in this case, given these particular facts, the in-court identification cannot stand without the out-of-court identification.
We are acutely aware that the conclusion that we feel we must reach will cause further reliving of the pain and uncertainty for the Epperson family who are entitled to closure and justice. Justice demands that the guilty [331]*331be punished, and the innocent set free. But the determination of who is guilty, and who is not guilty must be decided by those elemental rules of fairness set forth in our Constitution, as interpreted by the Supreme Court, and our own Court of Criminal Appeals. The jury had a difficult time with this case. First they told the judge they were deadlocked. Then, after two-days deliberation, they did reach a verdict. Their assessment of such a minimal punishment of 25 years for this brutal crime, with no mitigating factors, however, reflects much uncertainty. It is not this court’s job to find guilt or innocence, and we do not do so. It is our job to insure that evidence that is admitted to the jury is properly admitted, and insofar as the in-court identification, we don’t think it was. Sometimes an in-court identification is nothing more than a formality. This is obviously not that kind of case.
We sustain the defendant’s third point of error. In line with this decision, we do not need to reach the second point of error. We reverse and remand for a new trial.
DUNCAN, J., concurs in the judgment only.