Separate Statement of
HARRY T. EDWARDS, Circuit Judge:Though he concurs in the foregoing Opinion for the Court, Judge MacKinnon has written separately,1 expressing his individual view that there may not have been adequate evidence presented at trial to support the verdicts rendered. Though, before the District Court, appellants routinely moved for judgment notwithstanding the verdict, they have not argued on appeal the question of the sufficiency of the evidence. In these circumstances, it seems to me not only unnecessary but unwise for this court to discuss the adequacy of the evidentiary foundation for the decisions reached below. My reluctance to engage in such an inquiry is overcome, however, by my sense of the importance of not allowing some of the statements in the Concurring Opinion to pass uncontested. I, therefore, offer my own separate statement in order to prevent the creation of any misimpressions regarding the content of the legal doctrines governing permissible conditions of incarceration and to dispel any doubts concerning the sufficiency of the evidence adduced below.
I. Prison Conditions As “Cruel and Unusual Punishment”
In my view, the most serious misstatements of law in the Concurring Opinion concern standards of liability under the Eighth Amendment. Four aspects of its presentation of the constitutional doctrine are especially problematic: its explication of the “totality of the circumstances” test; its discussion of permissible levels of prison violence; its analysis of the effect on con*957stitutional norms of invocation of the class action procedure; and its assessment of the significance of “average” prison conditions in determining what constitutes “cruel and unusual punishment.” These issues are discussed in order below.
A. The “Totality of the Circumstances” Test
One of the principal arguments used by the Concurring Opinion to call into question the sufficiency of the evidence is that ap-pellees’ allegations of constitutional infringement were limited, for the most part, to one aspect or dimension of their incarceration. The Concurring Opinion insists that the inmates alleged only that they were exposed to excessive risk of assault and sexual abuse (caused or aggravated by various conditions), and that appellees failed to show or even assert that other conditions at Maximum — such as health care, food service and rehabilitative and recreational programs — fell below constitutional standards. The Concurring Opinion thus suggests that, because appellees’ complaint was so narrowly focused, they were not entitled to relief unless they could show that conditions in the identified dimension were unusually brutal.
In advancing this argument, the Concurring Opinion intends to rely on a legal theory that has been adopted in many judicial opinions construing the Eighth Amendment — a theory frequently referred to as the “totality of the circumstances” test.2 The suggestion in the Concurring Opinion that one must consider the “totality of the circumstances” at a given prison, by itself, surely is not objectionable. Rather, what is troublesome here is the apparent construction of the “totality of the circumstances” test offered in the Concurring Opinion — a construction that I find at odds with judicial precedent.
Hitherto, the “totality of the circumstances” test always has been used as shorthand for one or both of two insights. First, separate prison conditions that exacerbate one another must, of course, be considered together to determine whether their cumulative impact is sufficient to violate the ban on “cruel and unusual punishment.” Thus, when a particular procedure or form of treatment, such as punitive or solitary confinement, is challenged, one must examine all of the constituent aspects of that form of incarceration to determine whether, in combination, they run afoul of the Constitution. Hutto v. Finney, 437 U.S. 678, 685-88, 98 S.Ct. 2565, 2570-72, 57 L.Ed.2d 522 (1978); Maxwell v. Mason, 668 F.2d 361, 363 & n. 8 (8th Cir.1981). The same principle applies when one is assessing the constitutional status of a prison as a whole. Insofar as “[t]he ‘touchstone’ of the Eighth Amendment inquiry is ‘the effect upon the imprisoned,’ ” Rhodes v. Chapman, 452 U.S. 337, 366,101 S.Ct. 2392, 2409, 69 L.Ed.2d 59 (1981) (Brennan, J., concurring) (quoting Laaman v. Helgemoe, 437 F.Supp. 269, 323 (D.N.H.1977)),3 it is not sufficient to consider, seriatim, the influence on the prisoners of each of various arbitrarily designated dimensions of prison life (health care, food, inmate safety, etc.). Conditions in each of these areas “exist in combination; each affects the other; and taken together they [may] have a cumulative impact on the inmates .... ” Holt v. Sarver, 309 F.Supp. 362, .373 (E.D.Ark.1970), aff’d, 442 F.2d 304 (8th Cir.1971). “Even if no single condition of confinement would be unconstitutional in itself, ‘exposure to the cumulative effect of prison conditions may subject inmates to cruel and unusual punishment.’ ” Rhodes v. Chapman, 452 U.S. at 363, 101 S.Ct. at 2407 (Brennan, J., concurring) (quoting Laaman v. Helgemoe, 437 F.Supp. at 322-23)4
*958The second insight embodied in the “totality of the circumstances” test consists of a recognition of the unusual structure and purpose of most suits challenging terms of incarceration on the basis of the Eighth Amendment. Such controversies belong to a special category of litigation that has been denominated “complex enforcement.” 5 Among the atypical features of suits of this sort, two are relevant in the present context. First, the wrong attacked is not a single, discrete transgression on the part of the defendant, but a system of related behavior, a “practice” engaged in over time by an entire institution.6 Second, the willingness (on the part of both plaintiff and court) to mount such a systemic attack reflects a “transformative vision,” a conviction that the real source of illegality is the organization of the institution in question, not the delicts of its individual officials, and that the law can and should dig deep in and, if necessary, substantially reorder that institution to ensure that its conduct is consistent with the society’s fundamental norms.7 As applied to Eighth Amendment suits, both of these features imply that the attention of the parties and the adjudicator ought to be directed at the structure and conduct of the prison as a whole, not at separate facets or conditions.8
To the extent that my colleague means to refer to these two insights, his invocation of the “totality of the circumstances” test is unexceptionable. But the manner in which the Concurring Opinion distinguishes the instant action from some prior cases involving broader spectra of constitutional violations suggests that more may be intended. It would appear from the Concurring Opinion that attention to the “totality of the circumstances” implies that a prison must be rife with brutality before constitutional barriers are breached. It is thus argued that one or a few “deplorable factors” rarely will be enough; inmates usually must demonstrate “a substantial number” of outrageous conditions before they can get relief.9 Neither of the two insights discussed above lends any support to this suggestion. That two or more conditions may exacerbate one another does not mean that a single condition, if serious enough, cannot violate the Eighth Amendment. And that an adjudicator often must enlarge his field of vision to include all aspects of an institution’s structure and behavior does not imply that he must be blind to one vicious or degrading practice.
In my view, the test proposed in the Concurring Opinion is also inconsistent with *959an enormous body of case law. The opinions are legion that indicate that unacceptable conditions in even one of the dimensions identified by the Concurring Opinion will run afoul of the Constitution.10 Moreover, the courts in most of the cases that have involved “a substantial number of deplorable factors” have stated the rules thereby violated as independent constitutional requirements — making it clear that they were identifying and condemning a series of separate Eighth Amendment violations, any one of which alone would constitute “cruel and unusual punishment.”11 Finally, the standard advocated by the Concurring Opinion is inconsistent with the Supreme Court’s recent analysis of Eighth Amendment doctrine. In Rhodes v. Chapman, the Court reaffirmed the holdings of Estelle v. Gamble and Hutto v. Finney, that either the denial of medical care by itself or the kind of multifaceted inhumanity that characterized the Arkansas prisons would constitute “cruel and unusual punishment.” 452 U.S. at 347, 101 S.Ct. at 2399. The Court then went on to say:
Conditions other than those in Gamble and Hutto, alone or in combination, may deprive inmates of the minimal civilized measure of life’s necessities. Such conditions could be cruel and unusual under the contemporary standard of decency that we recognized in Gamble.
Id. (emphasis added).12
The test elaborated in the Concurring Opinion appears especially implausible *960when applied to the facts of this case. Assuming, arguendo, that a poor diet or inadequate clothing, without more, cannot rise to the level of a constitutional violation, it is difficult to see how the same can be said of the prevalence of violence. Many members of the appellee class have been beaten, stabbed or raped.13 The remainder live in constant fear of such assaults. Undoubtedly several will be physically or psychologically maimed when — and if — they leave the prison. The suggestion that, unless coupled with other forms of inhumanity, prolonged subjection to such harms and risks cannot constitute “cruel and unusual punishment” is impossible to fathom.
In short, the Concurring Opinion appears to seize upon the concept of the “totality of the circumstances,” ignore its well-established meaning and legitimate foundations, and transform it into a strong presumption that prisoners must show that they have been brutalized by not one but “a substantial number” of inhumane conditions before they are entitled to relief under the Eighth Amendment. The only suggestions in the Concurring Opinion as to why such a drastic doctrinal reform might be warranted are the references to judicial language counseling caution and restraint when assessing prison conditions. Those admonitions, however, provide no support for the proposal advanced in the Concurring Opinion. While it is certainly true that federal courts must accord legislators and prison officials some latitude in designing and testing solutions “to the perplexing sociological problems of how best to achieve the goals of the penal function in the criminal justice system,” Rhodes v. Chapman, 452 U.S. at 352, 101 S.Ct. at 2402, it is equally true that the courts have a duty to protect the constitutional rights of prisoners.14 Every modern opinion in the field has affirmed as much. Thus, immediately following the language in Rhodes that the Concurring Opinion finds so telling, the Supreme Court makes it plain that:
Courts certainly have a responsibility to scrutinize claims of cruel and unusual confinement, and conditions in a number of prisons, especially older ones, have justly been described as “deplorable” and “sordid.” When conditions of confinement amount to cruel and unusual punishment, “federal courts will discharge their duty to protect constitutional rights.”
452 U.S. at 352, 101 S.Ct. at 2401 (footnote and citations omitted).15
B. Exposure to Violence as a Form of Punishment
Possibly because of the weak underpinnings of its variant of the “totality of the circumstances” standard, the Concurring Opinion does not place all of its weight on that stanchion. The opinion goes on to sketch the state of the law regarding permissible levels of inmate violence under the Eighth Amendment. The thrust of the analysis appears to be that, even if we consider only the violence issue, the appel-lees have failed to show that they were exposed to a sufficiently serious risk of harm to satisfy the stringent requirements for making out a constitutional violation.
The Concurring Opinion does not purport to state any clear legal standard concerning how prevalent prison violence must be before the Constitution is offended. However, various comments in the opinion — regarding the “unavoidability” of frequent *961assaults, the congenital viciousness of maximum-security inmates, and the legitimacy of prison officials taking “calculated risks” —give the general impression that nothing short of open warfare among the prisoners, countenanced by the guards and administrators, will state a claim under the Eighth Amendment.
To expose the fallacy inherent in this conception of the state of the law, one needs only to quote a few of the leading opinions in the field. The central norm in this area is that,
[a] prisoner has a right, secured by the eighth and fourteenth amendments, to be reasonably protected from constant threat of violence and sexual assault by his fellow inmates, and he need not wait until he is actually assaulted to obtain relief.
Woodhous v. Commonwealth of Virginia, 487 F.2d 889, 890 (4th Cir.1973) (per curiam) (citation omitted). Accord Ramos v. Lamm, 639 F.2d 559, 572 (10th Cir.1980), cert. denied, 450 U.S. 1041, 101 S.Ct. 1759, 68 L.Ed.2d 239 (1981); Withers v. Levine, 615 F.2d 158, 161 (4th Cir.), cert. denied, 449 U.S. 849, 101 S.Ct. 136, 66 L.Ed.2d 59 (1980); Little v. Walker, 552 F.2d 193, 197 (7th Cir.1977), cert. denied, 435 U.S. 932, 98 S.Ct. 1507, 55 L.Ed.2d 530 (1978); Doe v. Lally, 467 F.Supp. 1339, 1348, 1354 (D.Md.1979). Prison officials have a corresponding duty “to exercise reasonable care to prevent prisoners from intentionally inflicting harm or creating unreasonable risks of harm to other prisoners.” Withers v. Levine, 615 F.2d at 161. Accord Woodhous v. Virginia, 487 F.2d at 890; Doe v. Lally, 467 F.Supp. at 1354. Or, as the Ninth Circuit has recently put it, “[p]rison officials have a duty to take reasonable steps to protect inmates from physical abuse.” Hoptowit v. Ray, 682 F.2d 1237,1250 (9th Cir.1982) (citation omitted).
The opinion of the Fourth Circuit in Withers v. Levine persuasively elaborates these general principles. The Court of Appeals was concerned that a phrase used in its prior decision in Woodhous to describe conditions at the prison in question16 had been improperly construed as a statement of the threshold of liability. The court corrected the misinterpretation in the following terms:
A pervasive risk of harm may not ordinarily be shown by pointing to a single incident or isolated incidents, but it may be established by much less than proof of a reign of violence and terror in the particular institution. The defendants seized upon that explanatory phrase from Woodhous to contend that something approaching anarchy must be proven before a cause of action under Woodhous may be made out, but conditions need not deteriorate to that extent before the constitutional right to protection arises. It is enough that violence and sexual assaults occur on the idle tier at MHC with sufficient frequency that the younger prisoners, particularly those slightly built, are put in reasonable fear for their safety and to reasonably apprise prison officials of the existence of the problem and the need for protective measures.
615 F.2d at 161.
In sum, the Concurring Opinion holds ap-pellees to an unduly high standard of liability concerning impermissible levels of violence. To be entitled to relief under the Eighth Amendment, appellees need only have shown that violence and sexual assaults occurred at Maximum sufficiently often to induce prisoners reasonably to fear for their safety and to make the prison officials aware of the problem, and that those officials failed to take reasonable steps to protect the inmates.
C. Constitutional Norms and Class Actions
The Concurring Opinion attempts to buttress its novel formulations of the applicable standards of liability by attributing unusual significance to the fact that the case at bar was brought as a class action. The *962opinion is riddled with references to appel-lees’ status as a “class” but at no point does it clearly indicate why this seemingly irrelevant circumstance is so important. Various hints in the opinion suggest an adherence to one or both of two theories. The first is plausible but is not germane; the second is indefensible.
At a few points, the Concurring Opinion appears to refer to the fact that class action suits in which plaintiffs seek monetary damages for the deprivation of their civil rights are sometimes conducted in two stages. In the first stage, the issue of the defendant’s liability (j.e., the question whether the defendant has breached some statutory or constitutional duty it owes to all members of the plaintiff class) is adjudicated. In the second stage, which ensues only if such liability is established, the remedy to be awarded each individual plaintiff is determined. This second phase of the suit often entails the submission and evaluation of additional evidence — relating, for example, to the question whether the defendant’s misconduct was a “but-for” cause of the injury sustained by each claimant.17 If this is the scenario envisioned by my colleague, I have no quarrel with it;18 however, I fail to see its relevance to the question whether appellants breached their obligation not to subject appellees to cruel and unusual punishment. The fact that, under some conceivable procedure for processing the case, a finding of liability might be followed by the adducement of additional evidence pertaining to the relief to be granted individual prisoners is of no significance in the present context.
Other language in the Concurring Opinion seems to suggest not that class actions may be procedurally bifurcated in the fashion indicated, but that suits brought on behalf of a class should be subjected to more stringent substantive standards of liability than individual actions. This suggestion is unsupported and insupportable. The Concurring Opinion points to no case law or commentary adopting or advocating such an extraordinary theory. And no arguments, based on logic or policy, favor such a conception of the differences between the two forms of litigation.
Certainly the notion that individual and class actions are governed by different standards of liability makes no sense when applied to the facts of the case at bar. The heart of the appellees’ claim is that they have been and continue to be exposed to a risk of rape, assault or murder — a risk serious enough to give rise to a constitutional violation. All prisoners at Maximum have been and still are exposed to the same danger; any one of them could have brought an individual action challenging the conditions of his confinement. The fact that the inmates have combined their efforts and, with the permission of the District Court, have presented their claims in the form of a class action has no conceivable bearing on the merits of their allegations.
Nor does the theory proposed in the Concurring Opinion find any support in the voluminous case law dealing with Eighth Amendment challenges to prison conditions. The courts’ hitherto unquestioned belief that the standards of liability are the same in the two procedural contexts is manifested by the fact that they treat citations to opinions that involve class actions and citations to opinions that involve individual suits as interchangeable.19 Indeed, for any *963significant point of law in this field, one can find comparable authority in class action cases and in eases brought by individuals.20
The argument suggested in the Concurring Opinion appears even more fragile when one takes into account the principles upon which class actions are founded. Scholars continue to disagree over the best way to explain and justify the binding effect of judgments in such cases on ignorant or involuntary class members and the tendency of the class action device to facilitate the initiation of certain kinds of suits;21 however, all agree that the essence of the class action procedure is that “[i]t provides a means by which, where a large group of persons are interested in a matter, one or more may sue or be sued as representatives of the class without needing to join every member of the class.”22 In other words, a class action does not involve claims different in substance from those involved in an individual action. It is merely a mechanism for allowing a group of people, all of whom have suffered or are suffering the same wrong, to join together to seek redress.
This conclusion is reinforced by consideration of the status and function of Fed.R. Civ.P. 23, the provision on which class actions brought in the federal courts are based. Both the origins of Rule 23 and its interpretation in other contexts indicate that successful invocation of the provision does not trigger a shift in the applicable substantive standards of liability. The Enabling Act, pursuant to which the rule was promulgated, expressly provides that “[s]uch rules shall not abridge, enlarge or modify any substantive right .... ” 28 U.S.C. § 2072 (1976). Moreover, it is settled that Rule 23 is sufficiently confined to questions of “procedure” to be applicable in diversity actions without transgressing the constitutional requirement that federal courts defer to state “substantive” law in such situations. See 7 C. Wright & A. Miller, Federal Practice and Procedure § 1758 (1972); Briskin v. Glickman, 267 F.Supp. 600, 603-05 (S.D.N.Y.1967).23 In short, “Rule 23 is considered a procedural rule, rather than one that affects the substance or the merits of litigation.” H. New-berg, Class Actions § 2135 (1977) (footnote omitted).
Finally, consideration of the practical effect of the doctrine intimated in the Concurring Opinion makes its senselessness even more apparent. Were the proposed theory to be widely adopted, the prisoners in every future case, aware that they would be confronted with elevated standards of liability if they brought a class action, would likely instead bring a series of individual suits. The court would consequently be compelled to hear and evaluate a multiplicity of essential identical cases, rather than a single consolidated action. The resultant waste of time and energy is precisely what Rule 23 was designed to avoid.
In sum, the apparent suggestion in the Concurring Opinion that the appellees, because they brought their suit in the form of *964a class action, must satisfy specially stringent standards of liability, is unprecedented and untenable.
D. National Prison Standards and “Cruel and Unusual Punishment”
A significant portion of the hypothetical evaluation in the Concurring Opinion of appellees’ evidence takes the form of comparative analysis. Thus, in assessing appel-lees’ claim that the situation at the prison was aggravated by understaffing, the Concurring Opinion observes that the inmate-to-guard ratio at Maximum was better than the national average. In considering the prevalence of violence, the opinion stresses that assaults occurred less frequently at Maximum than at other prisons where Eighth Amendment violations have been found. Moreover, the opinion notes, the incidence of violence was within the range of figures that an expert witness had testified was to be “expected” in a prison like Maximum. The Concurring Opinion insists that these observations should be fatal to appellees’ Eighth Amendment cause of action. Unless inmates can show that the conditions of their confinement are worse than average, it argues, they cannot make out a claim that they have been subjected to “cruel and unusual punishment.”
When the Concurring Opinion applies this kind of comparative analysis to appellees’ negligence claims, it falls into patent error. It has long been hornbook law that conformity to industry standards is insufficient to establish due care. As Judge Learned Hand once put it:
Indeed in most cases reasonable prudence is in fact common prudence; but strictly it is never its measure; a whole calling may have unduly lagged in the adoption of new and available devices. It never may set its own tests, however persuasive be its usages. Courts must in the end say what is required; there are precautions so imperative that even their universal disregard will not excuse their omission.
The T.J. Hooper, 60 F.2d 737, 740 (2d Cir.) (citations omitted) (dicta), cert. denied, 287 U.S. 662, 53 S.Ct. 220, 77 L.Ed. 571 (1932).24
The mode of analysis employed in the Concurring Opinion is equally inappropriate when applied to the Eighth Amendment. Whatever may have been the original meaning of the constitutional provision,25 it is now well established that the ban on “cruel and unusual punishment” is defined with reference, not to prevailing penal practices, but to “the evolving standards of decency that mark the progress of a maturing society.” Estelle v. Gamble, 429 U.S. 97, 102, 97 S.Ct. 285, 290, 50 L.Ed.2d 251 (1976) (quoting Trop v. Dulles, 356 U.S. 86,101, 78 S.Ct. 590, 598, 2 L.Ed.2d 630 (1958)). In determining whether a given condition violates the amendment, courts and juries are to look, not to circumstances common in other prisons, but to “the public attitude toward a given sanction.” Rhodes v. Chapman, 452 U.S. at 349 n. 13,101 S.Ct. at 2400 n. 13 (1981) (quoting Gregg v. Georgia, 428 U.S. 153, 173,96 S.Ct. 2909, 2925,49 L.Ed.2d 859 (1976) (joint opinion)). The touchstone is “what the general public would consider decent.” Hoptowit v. Ray, 682 F.2d at 1246 (citation omitted).
Although the public at large may at times close its eyes to the sad truth, it is undeniable that conditions in the nation’s prisons often are extraordinarily brutal and degrading.26 It would thus not be surprising if many aspects of current “industry standards” fell below “what the general public would consider decent.” In any case, it is clearly erroneous to assume that “aver*965age” conditions will invariably pass constitutional muster.
II. The Sufficiency of the Evidence
As I indicated at the outset, this court should not- — indeed is not adequately prepared to — undertake a thorough examination of the sufficiency of the evidence to support the jury’s verdict. The following section does not purport to conduct such a comprehensive review of the record. Its purpose is merely to highlight the most serious errors in the survey of the testimony attempted in the Concurring Opinion.
A. The Standard of Review
The rule governing appellate review of the sufficiency of the evidence to support a jury verdict is both strict and well established. That standard requires that “the evidence, along with all inferences reasonably to be drawn therefrom, ... [be] viewed in the light most favorable to” the party in whose favor the verdict was rendered. Alden v. Providence Hospital, 382 F.2d 163,165 (D.C.Cir.1967). Moreover, unless an appellate court can determine that “no reasonable man could reach a verdict in favor of” the prevailing party, the jury’s verdict must be allowed to stand.27
Application of the foregoing standard of review to the facts of the instant case, even adopting the Concurring Opinion’s inflated liability theories, yields only one conceivable conclusion: we would be required to uphold the verdict if the strength of its evidentiary foundation were an issue before us. The most that could be said in appellants’ favor is that “reasonable men” might disagree over the sufficiency of the evidence; it surely could not be concluded, however, that “no reasonable man could reach a verdict in favor of” appellees. Only by miscon? struing statistical evidence and ignoring! a plethora of incriminating testimonial evidence adduced below, is the Concurring Opinion able to suggest that the matter is even worthy of discussion.
B. The Treatment of Statistical Evidence
Numerical evidence presented at trial, particularly that relating to levels of violence, is misused in the Concurring Opinion in three ways. First, the opinion contrasts, somewhat carelessly, the statistics relating to the incidence of assaults at Maximum with figures gathered at other institutions with only selective regard for the comparability of the respective reporting systems and prison populations. When a reference to comparability is advantageous, it is incorporated in the discussion. But when such favorable references are not available, the Concurring Opinion does not hesitate to evaluate violence statistics without any background information whatsoever.28 This skewed treatment of the data only compounds the problems created by reliance upon such comparative analysis in the first place.29
Second, the Concurring Opinion makes much of the fact that the average incidence of reported assaults at Maximum is “only fifteen” per year. But the figure most relevant to the appellees’ claims is not the average level of violence over the preceding seven and a half years, but the incidence of *966violence at the time of the suit.30 When that statistic is extracted from the record, the situation is revealed to be much more serious than the Concurring Opinion would have us believe. In the preceding six months there had been fifteen reported assaults at Maximum — an average of thirty per year — and the level of violence had been steadily rising since 1978. App. 157-58, 176.
An even more serious problem concerns the value of the statistics in question, no matter how they are construed, in assessing conditions at the prison. As testimony at the trial makes evident, very few of the assaults and homosexual rapes committed at Maximum ever find their way into the record books. Fear of reprisal, desire not to be known as a “snitch” and, in the case of homosexual attacks, reluctance to identify oneself as an “easy mark” deter victims from reporting such incidents.31 Under the circumstances, the fact that the prison has records of an inmate assault rate of “only” thirty per year proves little. Indeed, given the dynamics of the situation, it would seem likely that, as the level of violence and the danger of reprisal increase, the percentage of incidents reported would decrease, thus compounding the problem.
C. The Profusion of Testimonial Evidence
The crucial factor, for the purpose of assessing the appellees’ Eighth Amendment claim, is the seriousness of the risk of assault and sexual abuse to which they were exposed. Given the absence of reliable statistical indices of that risk, the trier of fact would have had to rely upon descriptions of the level of danger by inmates and prison officials. The record is replete with such testimony. Witness after witness described stabbings, beatings and fights. App. 264-65, 277, 294-97, 312,315,349,359, 362. Several testified that they had observed homosexual rapes. App. 283, 286-87, 291-93, 338, 373. The more detailed accounts included descriptions of: a stabbing in the shower area, App. 274-75, 284-85, an attempted stabbing with a butcher knife, App. 251-52, beatings in the shower area, App. 280, cell burnings, App. 322-323, 363, an inmate “bleeding profusely from his stomach,” App. 252, 271, an inmate thought to be a “snitch,” picking himself up off the floor, bleeding from the mouth and nose, App. 266-69, a stabbed inmate bleeding from the chest, App. 339, and the death of an inmate resulting from seven or eight stab wounds near his heart, App. 348. To suggest that no reasonable man, presented with this evidence, could conclude that prisoners at Maximum were exposed to a pervasive risk of harm reflects both contempt for the judgment of the trier of fact32 and callousness toward the appellees’ plight.
III. Conclusion
In light of the foregoing discussion, appellants’ decision not to raise on appeal the *967question of the sufficiency of the evidence is readily understandable. When the evidence adduced below, viewed dispassionately and through the lens of the applicable standard of appellate review, is measured against well-established substantive standards of liability under the Eighth Amendment, it becomes clear that the jury’s verdict is not vulnerable to reasoned criticism.
Separate Statement of
ROBB, Senior Circuit Judge:Although I share Judge MacKinnon’s doubts that the plaintiffs made a case in the District Court, I acquiesce in Judge Edwards’ opinion for the Court.
. For the sake of convenience, I will refer to my colleague’s separate statement as the “Concurring Opinion.”
. See, e.g., Clay v. Miller, 626 F.2d 345, 347 (4th Cir.1980) (per curiam).
. Cf. Carey v. Piphus, 435 U.S. 247, 254, 98 S.Ct. 1042, 1047, 55 L.Ed.2d 252 (1978) (reaffirming the principle that the purpose of constitutional tort actions is “to compensate persons for injuries caused by the deprivation of constitutional rights”) (discussed in the Opinion for the Court, Part IV.).
. See also Smith v. Sullivan, 553 F.2d 373, 378 (5th Cir.1977); Williams v. Edwards, 547 F.2d 1206, 1211, 1219 (5th Cir.1977). But see Hoptowit v. Ray, 682 F.2d 1237, 1246-47 (9th Cir.1982); Wright v. Rushen, 642 F.2d 1129, 1132-*95833 (9th Cir.1981) (both holding that the conditions in the various dimensions of prison life must be evaluated independently). Without approving of the holdings of Hoptowit and Wright, it is worth noting that they are incompatible with my colleague’s conclusion that the coalescence of “a substantial number of deplorable factors” is necessary to give rise to cruel and unusual punishment.
. See Note, Complex Enforcement: Unconstitutional Prison Conditions, 94 Harv.L.Rev. 626, 626 (1981); L. Sargentich, Complex Enforcement (March 1978) (unpublished manuscript) (on file at the Harvard Law School Library). Other studies of this special kind of litigation include Chayes, The Role of the Judge in Public Law Litigation, 89 Harv.L.Rev. 1281 (1976); Ei-senberg & Yeazell, The Ordinary and the Extraordinary in Institutional Litigation, 93 Harv.L. Rev. 465 (1980).
. Sargentich, supra note 5, at 9-22.
. Id. at 77-85.
. Only relatively recently has the Eighth Amendment begun to be used in this systemic way. Before the 1960s, it was principally used to evaluate individual, judicially-imposed sentences. See Note, Creatures, Persons, and Prisoners: Evaluating Prison Conditions Under the Eighth Amendment, 55 S.Cal.L.Rev. 1099, 1101-02 (1982). The modern usage of the provision, however, is now firmly established and has received the imprimatur of the Supreme Court. See Rhodes v. Chapman, 452 U.S. at 345-47, 101 S.Ct. at 2398-99.
. In fairness, I would note that, at a few points, the Concurring Opinion appears to concede that “a single factor” might give rise to a constitutional violation — at least if “so pervasive as to constitute a systemic infliction.” To the extent that these concessions represent acknowledgments that the “totality of conditions” test does nothing more (and nothing less) than embody the two insights just described, I cease to quarrel with this aspect of my colleague’s opinion. To the extent that the retreat is incomplete, my objection stands.
. See, e.g., Estelle v. Gamble, 429 U.S. 97, 103-04, 97 S.Ct. 285, 290-91, 50 L.Ed.2d 251 (1976) (dicta) (medical care); Maxwell v. Mason, 668 F.2d at 363-65 (conditions in solitary confinement); Withers v. Levine, 615 F.2d 158, 161 (4th Cir.) (threat of sexual assault from fellow inmates), cert. denied, 449 U.S. 849, 101 S.Ct. 136, 66 L.Ed.2d 59 (1980); Leonardo v. Moran, 611 F.2d 397, 398-99 (1st Cir.1979) (dicta) (threat of violence from other inmates); Burks v. Teasdale, 603 F.2d 59, 62-63 (8th Cir.1979) (overcrowding); Todaro v. Ward, 565 F.2d 48, 52-53 (2d Cir.1977) (medical care); Little v. Walker, 552 F.2d 193, 197 (7th Cir.1977) (exposure to violence and sexual abuse by other inmates), cert. denied, 435 U.S. 932, 98 S.Ct. 1507, 55 L.Ed.2d 530 (1978); Woodhous v. Commonwealth of Virginia, 487 F.2d 889, 890 (4th Cir.1973) (per curiam) (threat of violence and sexual assault from other inmates); Stewart v. Rhodes, 473 F.Supp. 1185, 1193 (S.D.Ohio, E.D.1979) (prolonged, punitive use of “four-way” physical restraints violates the Eighth Amendment [a holding unrelated to the court’s additional finding that racial segregation of prisoners violates the Fourteenth Amendment]), appeal dismissed, 661 F.2d 934 (6th Cir.1981); Kish v. County of Milwaukee, 48 F.R.D. 102, 103-04 (E.D.Wis.1969) (exposure to violence and sexual abuse by other inmates), subsequent disposition aff'd, 441 F.2d 901 (7th Cir.1971).
. See, e.g., Hoptowit v. Ray, 682 F.2d at 1246-47, 1250, 1252-53, 1257-58, 1259; Wright v. Rushen, 642 F.2d at 1133 (dicta); Campbell v. Cauthron, 623 F.2d 503, 505-09 (8th Cir.1980); Williams v. Edwards, 547 F.2d at 1219 (after affirming the judgment of the trial court that “the totality of conditions” at the prison was unconstitutional, the court of appeals went on to hold that, by itself, “the level of medical care” at the prison “violated the eighth and fourteenth amendments”); Kirby v. Blackledge, 530 F.2d 583, 587 (4th Cir.1976) (holding that “[t]he combination of conditions alleged by the prisoners ... have the cumulative effect of being cruel and unusual punishment.. .. Many of the circumstances taken alone reach the level of cruel and unusual punishment, such as the Chinese cell, inadequate exercise and medical treatment, inadequate heating and ventilation, and lack of access to the prison library.”); Finney v. Arkansas Bd. of Correction, 505 F.2d 194, 200-09 (8th Cir.1974); Gates v. Collier, 501 F.2d 1291, 1301-10 (5th Cir.1974) (applying the cumulation theory to various factors affecting inmate safety but holding that plaintiffs had proved independent constitutional violations in the areas of hygiene, health care, solitary confinement, corporal punishment, and reliance upon “trusty” guards); Martinez Rodriguez v. Jimenez, 409 F.Supp. 582, 594 (D.P.R.1976), aff'd, 551 F.2d 877 (1st Cir.1977). But see Holt v. Sarver, 309 F.Supp. at 373.
. Justice Brennan, joined by Justice Blackmun in concurrence, laid more stress than did the majority on the need to consider the totality of prison conditions. 452 U.S. at 362-63, 101 S.Ct. at 2407. But he too recognized that a single circumstance of incarceration, inconsistent with contemporary standards of decency, would violate the Constitution. Thus he observed, “[e]ven if," in a given case, “no single condition of confinement would be unconstitutional in itself,” the combined effect of prison conditions may violate the Eighth Amendment — clearly leaving open the possibility that a single degrading circumstance might, by itself, amount to cruel and unusual punishment. Id. at 363, 101 S.Ct. at 2407 (emphasis added).
. See Part II.C. infra.
. Indeed, the special place of prisoners in our society makes them more dependent on judicial protection than perhaps any other group. Few minorities are so “discrete and insular,” so little able to defend their interests through participation in the political process, so vulnerable to oppression by an unsympathetic majority. Federal courts have a special responsibility to ensure that the members of such defenseless groups are not deprived of their constitutional rights. See United States v. Carolene Prods. Co., 304 U.S. 144, 153 n. 4, 58 S.Ct. 778, 784 n. 4, 82 L.Ed. 1234 (1938); J. Ely, Democracy and Distrust 135-36, 145-70 (1980).
. See also Bell v. Wolfish, 441 U.S. 520, 562, 99 S.Ct. 1861, 1886, 60 L.Ed.2d 447 (1979); Procunier v. Martinez, 416 U.S. 396, 405-06, 94 5. Ct. 1800, 1807-06, 40 L.Ed.2d 224 (1974); Cruz v. Beto, 405 U.S. 319, 321, 92 S.Ct. 1079, 1081, 31 L.Ed.2d 263 (1972) (per curiam).
. See Woodhous v. Virginia, 487 F.2d at 890 (“[C]onfinement in a prison where violence and terror reign is actionable.”).
. See, e.g., Baxter v. Savannah Sugar Ref. Corp., 495 F.2d 437, 444-45 (5th Cir.), cert. denied, 419 U.S. 1033, 95 S.Ct. 515, 42 L.Ed.2d 308 (1974).
. Indeed, the propriety of such a procedure was recently confirmed by this court. See Milton v. Weinberger, 696 F.2d 94, 99 (D.C.Cir.1982) (dicta).
. The single opinion that arguably might constitute an exception to this rule turns out, upon examination, to be ambiguous at best. In Maxwell v. Mason, 668 F.2d 361, 364-65 (8th Cir.1981), the Eighth Circuit distinguished the decision of the Fifth Circuit in Novak v. Beto, 453 F.2d 661 (1971), reh’g denied, 456 F.2d 1303, cert. denied, 409 U.S. 968, 93 S.Ct. 279, 34 L.Ed.2d 233 (1972), partly on the ground that “Novak was a class action in which the entire system of solitary confinement of the Texas Department of Corrections was challenged as unconstitutional.” 668 F.2d at 365. But the court immediately thereafter made clear that its principal reason for discounting the rele-*963vanee of Novak was that the latter involved a situation in which prisoners kept in solitary confinement were provided with clothing and blankets whereas the prisoner in the case before it had been given only undershorts and a mattress. 668 F.2d at 365. At no point did the Eighth Circuit suggest that Novak was inappo-site solely because it involved a class action rather than an individual suit.
. Thus, for example, of the ten cases cited in note 10, supra, in support of the principle that unacceptable conditions in even one of the dimensions delineated by the majority violate the Constitution, four involved class actions and six involved individual suits. Similarly, of the eight cases cited in note 11, supra, in support of the proposition that courts addressing multifarious violations most often state the constitutional norms thereby infringed as independent requirements, six involved class actions and two (at least at the stage of development at which they were being considered) involved individual suits.
. See Note, Developments in the Law — Class Actions, 89 Harv.L.Rev. 1318, 1337-43, 1348-72 (1976) (discussing the theories that currently seem most plausible).
. C. Wright, Handbook of the Law of Federal Courts 345 (1976).
. Briskin involved pendent, not diversity jurisdiction, but the court held that the same principles apply in the two contexts. 267 F.Supp. at 603.
. See also Texas & P. Ry. Co. v. Behymer, 189 U.S. 468, 470, 23 S.Ct. 622, 623, 47 L.Ed. 905 (1903); Morrison v. Kansas City Coca-Cola Bottling Co., 175 Kan. 212, 263 P.2d 217, 224 (1953).
. See Granucci, "Nor Cruel and Unusual Punishments Infíicted:" The Original Meaning, 57 Calif.L.Rev. 839 (1969).
. See generally Rhodes v. Chapman, 452 U.S. at 354-56, 101 S.Ct. at 2403-04 (Brennan, J., concurring); National Advisory Commission on Criminal Justice Standards and Goals, Corrections (1973); Gettinger, Accreditation On Trial, Corrections Mag., Feb. 1982, at 7; Andersen^ What Are Prisons For?, Time, Sept. 13, 1982, at 38.
. See, e.g., Muldrow v. Daly, 329 F.2d 886, 888 (D.C.Cir.1964). Several hundred cases adopting the same test are cited in 5A J. Moore & J. Lucas, Moore’s Federal Practice 50.02[1] nn. 20-21 (2d ed. 1982). The Concurring Opinion at a few points appears to suggest that the verdict should be overturned on appeal if it is not supported by “substantial evidence.” To the extent that those suggestions are intended to dilute the stringent “reasonable man” standard, they find no support in the voluminous case law.
. See, e.g., the references to the incidence of “brutality” and “assault” in the Duval County Jail, Jacksonville, Florida. The Concurring Opinion fails to consider whether the reporting systems are comparable and whether the statistics in question include minor altercations. (The opinion from which the data is drawn, Miller v. Carson, 401 F.Supp. 835, 883 (M.D.Fla.1975), strongly suggests, at a minimum, that the figures relate to more than serious assaults, as the Concurring Opinion implies.)
. Those difficulties are discussed in Part I.D. supra.
. Certainly, for the purposes of determining the existence of a constitutional violation, the relevant variable is the incidence of violence at the time of the suit. The measure or scope of the remedy to which the inmates — individually or collectively — are entitled may well turn partially on other factors, such as fluctuations in the incidence of violence in the course of their periods of incarceration. But the latter question is not before us.
. The affidavit of Deborah Jones, a correctional officer, dramatically illustrates this state of affairs:
I must emphasize that most inmate assaults go unreported because of the inmates’ realistic fear of retaliation by other inmates. Burning of cells is an increasingly common form of retaliation. Understaffing prevents the correctional staff from effectively dealing with this grave security problem. Indeed, just four weeks ago, a cell burning occurred in cellblock 2; the fire was so intense and the destruction so complete that the inmate’s typewriter melted.
App. 178. See also App. 265-69, 321-23. The reluctance of prisoners to report incidents— even when under oath at trial — is evident from the testimony of Mr. Davenport. App. 286-93, 295-99. Similar conditions are documented in Doe v. Lally, 467 F.Supp. at 1348 — 49; Withers v. Levine, 449 F.Supp. 473, 475, 478 (D.Md.1978), aff'd, 615 F.2d 158 (4th Cir.), cert. denied, 449 U.S. 849, 101 S.Ct. 136, 66 L.Ed.2d 59 (1980).
. The Concurring Opinion’s disdain for the ability of the trier of fact to evaluate the evidence is further manifested by its reference to the paucity of “jury expertise.”