Separate Statement of
MacKINNON, Circuit Judge:
For the reasons set forth in the opinion for the court, the judgment of the district court is vacated and the case remanded for a new trial. I am moved to point out, however, that I acquiesce in the court’s decision only because defendants have not appealed from the district court’s decision to deny their motion for judgment n.o.v. I harbor grave doubts whether plaintiffs have adduced sufficient evidence to support a jury verdict. It is my hope that the district court will conduct the new trial with an eye to the concerns expressed below.
A class action brought by all the prisoners in the Maximum Security Complex (Maximum) at the District of Columbia Reformatory at Lorton, Virginia (Lorton) against the District of Columbia municipal and reformatory officials (officials) who administer the penal institution,1 resulted in a judgment awarding money damages to the class and granting an injunction. Defendants were sued in their official capacities and the theory of plaintiffs’ case required them to prove that the D.C. defendants were liable because of acts resulting from official policy. Cf. Monell v. New York Department of Social Services, 436 U.S. 658, 690-92, 98 S.Ct. 2018, 2035-36, 56 L.Ed.2d 611 (1978). The jury award of damages and the court order of injunctive relief granted to the class of prisoners were (1) for the alleged infliction by the officials of “cruel and unusual punishment” (a constitutional tort)2 by the District of Columbia officials responsible for the administration of Maximum and (2) for the alleged negligent failure of the defendant officials to provide adequate protection against inmate assault. Each of the six counts of the complaint is based on violence, i.e., the officials’ “failure to provide adequate security and protection has caused plaintiffs to be exposed to pain and suffering by the constant risk and likelihood of physical violence .... ” Amended Complaint H90, JA 33 (emphasis added). The class was certified as all prisoners inside Maximum and the “common questions of law and fact affecting the rights of inmates [was stated as the right] to be free from actual and threatened inmate violence and pervasive risk of harm.” Amended Complaint ¶7, JA 15 (emphasis added). The D.C. and reformatory officials appeal the judgment on a variety of procedural bases and seek remand for a new trial.
[949]*949Basically my review of the complete trial record left me unconvinced that the finding of a constitutional tort and negligence by the officials was supported by substantial evidence of sufficient magnitude to support a class action. It was also a matter of concern to me that damages in a substantial amount, over $500,000 excluding interest, were to be paid to the class, i.e., all prisoners in Maximum. The plaintiff prisoners, who claim to be representative of the class, are confined for the following offenses: Two for first degree murder; one for murder, kidnapping and armed robbery; two for second degree murder; one for second degree murder and bank robbery; one for manslaughter; six for armed robbery; one for armed robbery and attempted rape; one for assault with a dangerous weapon; and one for grand larceny. Amended Complaint ¶¶ 12-28, JA 16-18.
Judge Edwards’ opinion for the court, at pages 953-956, points out the error in the court’s instruction that led to the monetary award. It is also significant that the class included all the prisoners who were creating the conditions of which they complained. Appellants have correctly noted on the record that
[p]laintiffs were ... both assailants and victims, weapon possessors, violators of security rules, men always ready to take advantage of one another by violent means and totally unwilling to share knowledge of weapons or assaults with prison personnel ....3
The prisoners demanded damages and in-junctive relief from D.C. officials and prison administrators for alleged violence that they refused to report (Tr. 265, 280, 294, 297, 322, 338, 349, 375). More significantly, this case would represent the first case in America that diligent research could find where an award of money damages to a class of prisoners had been upheld. Individual prisoners had previously been awarded money damages for individual torts, but so far as I can find, money damages have never been paid to an entire class of prisoners despite several cases of very egregious prison conditions. And the evidentiary record here indicates that conditions at Lorton do not even approach the conditions that were present in cases like Holt v. Sarver, 309 F.Supp. 362 (E.D.Ark.1970), aff’d, 442 F.2d 304 (8th Cir.1971), involving the Arkansas State Penitentiary System; Gates v. Collier, 501 F.2d 1291 (5th Cir.1974), involving the State Penitentiary at Parchman, Mississippi; or the Alabama prisons which were the subject of the decisions in Pugh v. Locke, 406 F.Supp. 318 (M.D.Ala.1976), aff’d as modified sub nom. Newman v. Alabama, 559 F.2d 283 (5th Cir.1977), rev’d in part sub nom. Alabama v. Pugh, 438 U.S. 781, 98 S.Ct. 3057, 57 L.Ed.2d 1114 (1978) (per curiam), where injunctions were issued to correct conditions in penitentiaries. It is incongruous to pay money damages to prisoners who are causing the violence and unhygienic conditions, who adhere to a code of silence, and who refuse to report violations or to cooperate (Tr. 287-89) in ameliorating the conditions about which they complain.
Typical of conditions in the prisons involved in the above cited cases, where complainants were limited to injunctive relief, were the conditions in the Arkansas State Penitentiary System. The Eighth Circuit decision found that prison to be fraught with cruel and unusual conditions, including, inter alia, the following: Ninety-nine percent of the security force at one of the prisons was comprised of so-called “trusty inmates” or “trusties” — privileged inmates who lorded over, abused, tortured, beat, and even occasionally murdered fellow inmates under their charge. Holt v. Sarver, supra, 309 F.Supp. at 373-76. With respect to the “trusty” system, the court noted that “just about every abuse which the system is capable of producing has been produced and is being practiced in this State.” Id. at 374. [950]*950Because of the lack of qualified inmate supervision — one facility housing 1,000 inmates had eight noninmate guards — violence among inmates went completely unchecked. This was particularly true in the “open barrack” dormitories, where upwards of 100 inmates per barrack were assigned “without regard to anything but rank and race.” Id. at 376. Stabbings in the dead of night by “crawlers” and “creepers”4 were commonplace and often condoned by the “trusties” themselves, who were often “ ‘in league with the assailants.’ ” Id. (quoting Holt v. Sarver (Holt I), 300 F.Supp. 825, 830 (E.D.Ark.1969)). Gang rapes were frequent, often with “trusties” “lookftng] on with indifference or satisfaction.” Id. at 377. Of this barrack-style confinement, the court said, in sum:
Sexual assaults, fights, and stabbings in the barracks put some inmates in such fear that it is not unusual for them to come to the front of the barracks and cling to the bars all night. That practice, which is of doubtful value is called “coming to the bars” or “grabbing the bars.” Clearly, a man who has clung to the bars all night is in poor condition to work the next day.
Id. Isolation cells were “overcrowded, filthy, and unsanitary.” Id. at 378. Inmates in isolation were given unsuitable food and were forced to live among rats. Id. For the inmates in general, there was no rehabilitation program of any kind. Id. at 378-79. Medical and dental care was subpar and largely unavailable. Id. at 380. Sanitary conditions in the kitchen were “deplorable.” Id. In sum, these conditions exhibited “a dark and evil world,” id. at 381, far beyond the constitutional pale. “However constitutionally tolerable the Arkansas system may have been in former years, it simply will not do today as the Twentieth Century goes into [its] eighth decade.” Id. at 381.
Other cases finding prison conditions unconstitutional address a concatenation of factors similar to those condemned in Holt, Gates, and Pugh.5 Usually, a substantial number of deplorable factors coalesce in prisons where a constitutional violation has been found. Where — as here — only a single factor is alleged — violence—it must be so pervasive as to constitute a systemic infliction, particularly in a class action. It is elementary that a judgment in a class action must be based on proof of harm to the class. In the present case, there is insufficient evidence viewed in the light most favorable to the verdict from which a reasonable juror could infer that the official policies of the defendants caused a systematic exposure of the class to an unreasonable risk of physical harm.
Judge Edwards’ separate statement expends its first nine pages misconstruing the foregoing paragraph, thereby setting up a straw man who would impose a “strong presumption that prisoners must show that they have been brutalized by not one but a ‘substantial number* of inhumane condi[951]*951tions before they are entitled to rélief under the Eighth Amendment,” p. 960 (emphasis added). But my opinion states that “a single factoi’" may be sufficient, supra at 950; it does not require “a substantial number of inhumane conditions.” So to assert is a gross misstatement.6 A single systemic condition can support a class action.
Class actions, as discussed at pages 961 to 964 of my colleague’s separate opinion, are designed to eliminate the necessity of bringing “a multiplicity of identical cases.” But class actions were not intended or designed to reduce the standard of proof that one single case required or to substitute a large smoke screen of facts for concise proof of the basic elements required to prove the case alleged. That is practically what I see happening here in the failure of the plaintiffs to prove, inter alia, that any official conduct proximately caused even one of them the harm they allege. This is in addition to their failure to prove under-staffing, excessive violence or overcrowding. See discussion, infra.
I. Violence
The Maximum prisoners complain of violence. However, this court and other courts have held that the existence of a number of assaults in a given prison is not per se “unreasonable” by common law negligence standards. Murphy v. United States, 653 F.2d 637 (D.C.Cir.1981). As we noted in Murphy, “[violence is unfortunately endemic to American prisons .... [A prisoner, and here the class, must prove] ... that [the] numbers [of assaults in question] were outside the range of violence normally associated with this type of penal institution, or that any subset of prisoners suffered from an unusual risk of attack.” Id. at 642 (emphasis added) (footnote omitted). Here plaintiffs were required to prove that the assaults clearly exceeded the range of violence normally associated with groups of the most violence-prone felons. Their evidence, viewed most favorably, does not constitute such proof. My colleague asserts that this court is not adequately prepared to undertake a thorough examination of the evidence to support the jury verdict. That statement is erroneous. The court is not only adequately prepared to thoroughly review the evidence, but that is its duty. And as a result of such review I find it insufficient to support the verdict.
Prisoners in Maximum are confined together because, of all prisoners in the institution, they have demonstrated by their prior conduct, in or out of prison, that they present the maximum risk of violence or escape. “Prison setting is, at best, tense. It is sometimes explosive, and always potentially dangerous.” Marchesani v. McCune, 531 F.2d 459, 462 (10th Cir.), cert. denied, 429 U.S. 846, 97 S.Ct. 127, 50 L.Ed.2d 117 (1976). The existence of some violence in prisons is unavoidable, given the character of the prisoners society must place behind bars.
As Justice Powell noted in Rhodes v. Chapman, 452 U.S. 337, 351, 101 S.Ct. 2392, 2401, 69 L.Ed.2d 59 (1981) (emphasis added) (footnote omitted):
This Court must proceed cautiously in making an Eighth Amendment judgment because, unless we reverse it, “[a] decision that a given punishment is impermissible under the Eighth Amendment cannot be reversed short of a constitutional amendment,” and thus “[revisions cannot be made in the light of further experience.” Gregg v. Georgia, 428 U.S. [153] at 176 [96 S.Ct. 2909, 2926, 49 L.Ed.2d 859], In assessing claims that conditions of confinement are cruel and unusual, courts must bear in mind that their inquiries “spring from constitutional requirements and that judicial answers to them must reflect that fact rather than a court’s idea of how best to operate a detention facility.” Bell v. Wolfish, 441 U.S. [520] at 539 [99 S.Ct. 1861, 1874, 60 L.Ed.2d 447].7
[952]*952And jury expertise is even more wanting than judicial expertise. Justice Powell went on in Rhodes to quote Procunier v. Martinez, 416 U.S. 396, 404-05, 94 S.Ct. 1800, 1807, 40 L.Ed.2d 224 (1974) (footnote omitted):
“[T]he problems of prisons in America are complex and intractable, and, more to the point, they are not readily susceptible of resolution by decree. Most require expertise, comprehensive planning, and the commitment of resources, all of which are peculiarly within the province of the legislative and executive branches of government. For all of those reasons, courts are ill equipped to deal with the increasingly urgent problems of prison administration and reform. Judicial recognition of that fact reflects no more than a healthy sense of realism.”
Rhodes v. Chapman, supra, 452 U.S. at 351 n. 16, 101 S.Ct. at 2401 n. 16.
Though Maximum does have some violence, the statistics offered by the prisoners at trial show an average of only fifteen assaults per year by inmates on other inmates from 1973 to 1980. JA 154-58. Such a low level of assaults does not support a claim of the infliction of cruel and unusual punishment upon plaintiffs under their own expert testimony, or under our decision in Murphy, which should control here — by virtue of its precedential value and its reasonableness. Comparative statistics were not introduced into evidence and so we are left to glean such statistics as we can from decided cases. First, the assaults in Maximum are very substantially lower than in Holt, Pugh, and the rest. In this case, while an average of fourteen (13.866) assaults per year by inmates on other inmates was reported for the seven-and-a-half year period from 1973 to July 1980,8 only one death may have occurred, and the evidence is inconclusive that it occurred within the period covered by the complaint. (Tr. 348). No riots causing mass injury took place. There was no “trusty” system, with its attendant ill-effects, in operation. In Murphy v. United States, supra, which involved the Lorton Youth Center — housing about 560 decidedly less violent inmates — this court ruled that twenty inmate assaults in the calendar year 1976, six of which occurred in a dormitory housing a hundred youths, were “not so high as to strike this court as per se unreasonable.” 653 F.2d at 642 n. 19 (emphasis added). In view of that precedent, with which my colleague fails to deal, I cannot conclude that an average of fourteen or fifteen assaults per year in the maximum security facility at issue in this case — which houses 400 inmates — is sufficient to support a finding that the number of assaults was unreasonable, excessive or indicative of insufficient protection. The nature of the population in Maximum alone would lead anyone to expect a greater risk of assault than would the population of the Youth Correction facility that was involved in Murphy. A comparison of the assault statistics for Lorton’s Maximum with those in Lorton’s Youth Correction complex is set [953]*953forth in the margin.9 Since the gravamen of the prisoners’ complaint is excessive violence, their proof does not support their complaint and their claim that the incidence of assaults was unreasonable cannot withstand our decision in Murphy.
On the issue of what amount of violence may reasonably be expected, it is also significant that Eugene Miller, the plaintiffs’ own penological expert witness, testified that twelve to twenty inmate assaults per year at Maximum was the expected range of such assaults. JA 243 — 44. Thus, by plaintiffs’ own standard, an average of fourteen or fifteen assaults per year, falling within that expected range and below that we found to be reasonable in Murphy, cannot be held to constitute sufficient evidence of the official infliction by defendants of cruel and unusual punishment on the class.
Prisoners with the most extreme records of violent propensities and violent crimes cannot expect that life in a maximum security prison, where their classification requires them to be housed, will be devoid of all violence. Such expectation is plainly unreasonable, except in an even more highly structured environment.10 All that the prisoners had a right reasonably to demand was that violence not be excessive. Their own evidence demonstrates that it is not.
In Part II.C. of his separate statement, my colleague refers to some of the evidence of violence at Maximum. While these incidents are horrendous, the character of the violence does not exceed what one might expect in a maximum security prison, and the .total number of assaults, as discussed fully elsewhere, is not excessive. As Judge Merhige stated in Penn v. Oliver, 351 F.Supp. 1292, 1294 (E.D.Va.1972):
It would be fantasy to believe that even the most enlightened prison officials operating with unlimited resources could prevent all acts of violence within the prison. Moreover, even if a prison official fails through his negligence to prevent an act of violence, a violation of constitutional right is not of necessity stated.
That case involved an individual complaint under 42 U.S.C. § 1983, and the principle is even more applicable to a class action case. My colleague’s opinion naively ignores the wisdom expressed by Judge Merhige. It should also be pointed out that the prisoners’ testimony on violence lacked specificity as to the parties involved and the dates. But the greatest defect in their testimony lies in the failure to establish a nexus between the unspecific assaults and the official policies of District officials — a failure to prove that official policy was the proximate cause of the harm complained of. While the testimony as to the assaults had the ring of truth for a prison such as Maximum, much of the testimony as to causation had a self-serving defect. On the general issue of the amount of violence and the staffing required, my colleague’s separate opinion misstates my view by asserting it to be my opinion that “average conditions” will invariably pass constitutional muster. But that is not my stated opinion, which is, [954]*954rather, that conditions much better than average — better, in fact, than those this court found reasonable in Murphy and which plaintiffs’ expert witness did not find to be unreasonable — were insufficient to support a finding of a constitutional violation.
II. Staffing
The prisoners also contend that staffing was inadequate and was thereby a cause of violence. In my view the prisoners have not discharged their burden to prove by a preponderance of the evidence that staffing was inadequate. Understaffing has been discussed extensively in several cases. The national average of inmates to guards in United States prisons is 5 to 1. Ruiz v. Estelle, 503 F.Supp. 1265, 1290 (S.D.Texas 1980), aff’d in part and rev’d in part, 679 F.2d 1115 (5th Cir.), amended in part and vacated in part, 688 F.2d 266 (5th Cir.1982). Any figure appreciably above that ratio has been held contributory to cruel and unusual prison conditions. In Ruiz v. Estelle, supra, 679 F.2d at 1140, for example, the Fifth Circuit determined that Texas state prisons had an unacceptably high inmate-to-guard ratio of 12.45 to 1. In Pugh v. Locke, supra, 406 F.Supp. at 322, 325, the ratio was 9.3 to 1 — still grossly inadequate.
At Lorton there were 122 guards at Maximum in charge of some 400 prisoners. JA 151. The inmate-to-guard ratio was thus 3.27 to 1. This is much better than the national average of 5 to 1, not to mention the ratios in Ruiz and Pugh. As a matter of law, therefore, such a ratio was not “woefully inadequate” or so “impermissibly high” as to constitute sufficient evidence that constitutional standards were violated. My colleague seeks to term this an argument for an “industry standard,” with its well-known defect, but the Maximum figure is 34 percent better than what he would term the industry standard. And my colleague does not suggest any ratio he would find acceptable. His argument is totally lacking in specificity. Plaintiffs’ evidence on understaffing, on which it bears the burden of proof, is obviously insufficient. Moreover, neither the prisoners nor my colleague have proven what the minimum standard is, and merely calling Maximum “understaffed” does not make it so.
Under D.C.Code § 24-442, the District of Columbia owes prisoners the identical duty of “reasonable care in the protection and safekeeping” owed them under principles of common law negligence. Gaither v. District of Columbia, 333 A.2d 57, 60 (D.C.App.1975). Prison personnel are not, however, insurers of an inmate’s safety. Matthews v. District of Columbia, 387 A.2d 731 (D.C.App.1978). Thus, the fact that an inmate is assaulted and sustains injuries does not, by itself, establish liability. District of Columbia v. Davis, 386 A.2d 1195, 1200 (D.C.App.1978). “The [inmates] must establish by competent evidence a standard of care; that the defendants] violated that standard; and that such violation proximately caused injury to the [inmates].” Hughes v. District of Columbia, 425 A.2d 1299, 1302 (D.C.App.1981). The prisoners’ evidence is insufficient to satisfy this standard.
The duty imposed on defendants requires “the exercise of ordinary diligence to keep prisoners safe and free from harm.” Jones v. United States, 534 F.2d 53, 54 (5th Cir.), cert. denied, 429 U.S. 978, 97 S.Ct. 487, 50 L.Ed.2d 586 (1976). Some antisocial behavior is to be expected in a prison environment; accordingly, the government is not an insurer of the safety of a prisoner, id.; Fleishour v. United States, 244 F.Supp. 762, 767-68 (N.D.Ill.1965), aff’d, 365 F.2d 126 (7th Cir.), cert. denied, 385 U.S. 987, 87 S.Ct. 597, 17 L.Ed.2d 448 (1966), or here, of the class of all prisoners in Maximum. “The government’s liability ultimately depends upon whether the risks it took were reasonable or unreasonable.” Cowart v. United States, 617 F.2d 112, 116 (5th Cir.) (emphasis added), cert. denied, 449 U.S. 903, 101 [955]*955S.Ct. 275, 66 L.Ed.2d 134 (1980).11 To impose liability on the District and its officials also requires proof that their actions were the proximate and foreseeable cause of the alleged injuries. Id. As stated above, this necessary element of the case has not been proved.
Courts will not readily find negligence (or the official infliction of cruel and unusual punishment) from the presence of some risk, as is indicated by the so-called “calculated risks” test first articulated in Fleish-our. There, the court recognized that it is the
standard practice in modern penal institutions to take calculated risks in various aspects of prison life, housing, work, recreation, religious worship and others, so that prisoners may learn to get along with other persons as part of the rehabilitation process. The result is, that from time to time, there occurs an episode such as that here involved in which the gamble is lost and another prisoner is injured.
244 F.Supp. at 767. See also Williams v. United States, supra, 384 F.Supp. at 584. Thus, in Fleishour, housing a number of inmates in one big room, a room in which there was a fire extinguisher on the wall, constituted a rehabilitative program that involved a calculated risk that excusably backfired. No finding of negligence was warranted — even in a case involving a single individual.
It is obvious from the record that the prison conditions here, to some extent, represent a calculated risk taken by preferring to allow more freedom rather than closer confinement. Were the officials negligent, or inflicting cruel and unusual punishment, in allowing the inmates at Maximum the freedom of movement that they enjoyed? If the result of allowing prisoners such freedom (with no overcrowding, a more than adequate staffing ratio of 3.27 to 1, and a not unacceptable prisoner assault average of fourteen or fifteen a year) were to be that District taxpayers must pay over a half a million dollars to the prisoners— many of whom cause the problems that exist — the eventual solution would probably be to increase greatly restrictions on prisoners’ freedom of movement by creating a more controlled environment. Locking prisoners in their cells for greater periods of time might be one solution. Also, their ability to associate with other prisoners might be substantially curtailed. Such practices might also impede rehabilitation. Would plaintiffs welcome a remedy that required them to be confined to their cells for more substantial periods of time, to have their freedom to associate with other prisoners more severely restricted, or to have their movements otherwise hobbled? If they prevail in their lawsuit, such may be the result.
Considering all the evidence, it is my view that the choices Lorton made were reasonable and were made in the prisoners’ interests. The result has been uncrowded living conditions, commendable freedom and a not unacceptable incidence of assaults. Prison .officials do not have the luxury of selecting their clientele. There will always be a danger of assaults when the most dangerous and violent criminals are confined in a single section of a prison facility, as they must be. But unless prisons lock up such prisoners in individual cells and greatly restrict their freedom of movement within the facility, as was once the practice, such institutions must be permitted to take calculated risks. In my view, the evidence here is insufficient to support a finding that the risks in Maximum exceeded what was reasonable or that the conditions at Lorton “ ‘involve[d] the unnecessary and wanton infliction of pain.’ ” Rhodes v. Chapman, supra, 452 U.S. at 346, 101 S.Ct. at 2398 (quoting Gregg v. Georgia, 428 U.S. 153, 173, 96 S.Ct. 2909, 2925, 49 L.Ed.2d 859 (1976) (Stewart, Powell, Stevens, JJ.)).
I also have some difficulty agreeing to a new trial. A new trial will not change the facts that average assaults were only fourteen or fifteen a year, that the staffing ratio was 3.27 to 1, or that 400 inmates in [956]*956Maximum do not establish overcrowding. Under the theory of the plaintiffs’ complaint, these unchangeable facts substantially undercut the principal basis for their claims that the class was harmed as a result, of pervasive negligence or that cruel and unusual punishment was officially inflicted on the class. It thus appears to me that unchangeable evidence as to vital elements of plaintiffs’ complaint disproves their claims. I also have considerable difficulty in supporting a conclusion that money damages can be awarded to a class of prisoners in mass that includes many prisoners who are causing the conditions complained of and will not cooperate to help correct them.12
I close with a summary statement regarding my colleague’s separate opinion. He attacks what statistics are available, but substitutes nothing in their place. He exaggerates the violence, understates the staffing, misstates my opinion, and deems it proper to reward a class of inmates for causing violence whose alleged level is based substantially upon speculation on unreported assaults. The number of definite assaults that were actually recounted at trial does not add up to a constitutional violation; to hold that it does would trivialize constitutional violations. My colleague’s reliance upon unreported assaults, and the assaults he assumes were committed after the period covered by the lawsuit, constitutes a tacit admission that the evidence of assaults in the record is insufficient to support the verdict.
However, as stated above, for reasons set forth in Judge Edwards’ opinion for the court, I concur in vacating the judgment and remanding the case for a new trial.
As he indicates in his separate statement, Judge Robb shares the doubts expressed above that plaintiffs made a case in the District Court, but acquiesces in Judge Edwards’ opinion for the Court.